Mechanisms of Action of Massage – Knowledge

Massage is among the most widely used interventions in both clinical care and competitive sport, yet a recurring conclusion across the scientific literature is that its mechanisms remain incompletely understood. Weerapong, Hume, and Kolt (2005) set out the classical framework — biomechanical, physiological, neurological and psychological mechanisms — while noting that most of the benefits attributed to massage rested on observation and tradition rather than on demonstrated physiological pathways. Two decades later, Dakić et al. (2023), synthesising 114 articles in a PRISMA-guided systematic review, reached a similar verdict: massage did not affect motor abilities with the exception of flexibility, and its value in sport appears to be largely indirect. This review organises the proposed mechanisms into five categories and rates each by the strength of supporting evidence.

A note on how Dakić et al. (2023) is used below, since it is the broadest source cited here. It is a narrative tabulation, not a meta-analysis: no effect sizes are pooled, no risk-of-bias or certainty assessment is reported, and all study designs except reviews were eligible, so uncontrolled and non-randomised work sits alongside RCTs. Findings are presented as counts of studies showing an increase, a decrease, or no effect — vote counting, which is insensitive to study size and quality. Two further features matter for interpretation: 48 of the 114 studies used a foam roller rather than manual massage, so several headline conclusions pool two distinct interventions (Davis et al., 2020, by contrast, excluded foam rolling by design); and the literature search was run in July 2020 despite publication in 2023. Its 2,731 participants exceed the 1,012 in Davis et al. (2020), but breadth of inclusion is not the same as strength of evidence. Dakić et al. is therefore cited here for the direction and consistency of findings across a wide literature, never as a quantitative estimate.

1. Biomechanical and Tissue Mechanisms

Status: Postulated — partial support (flexibility supported; adhesion and stiffness claims now meta-analytically null)

The oldest theory holds that manual pressure mechanically deforms tissue — stretching fascia, reducing adhesions, and altering muscle stiffness. Weerapong et al. (2005) framed this as the biomechanical model, proposing that mechanical pressure increases muscle compliance, thereby raising joint range of motion and lowering passive and active stiffness. Historically the idea drew on thixotropy (a gel-to-sol transition under mechanical stress), an explanation now largely regarded as outdated. The one motor outcome that improves across syntheses is flexibility: Dakić et al. (2023) found that massage did not alter motor abilities in general, with flexibility the notable exception, and Davis, Alabed, and Chico (2020) reported a statistically significant pooled effect (SMD 1.07, 95% CI 0.21 to 1.93; 7 studies, 246 participants) corresponding to a 7% increase in flexibility scores. That estimate is less secure than it looks: heterogeneity was very high (I² = 90%), the authors attribute the result to a single outlier study with the remainder showing a consistently smaller benefit, and every included trial compared massage against no intervention rather than against an alternative flexibility intervention. Notably, no human study has demonstrated a structural change in the muscle–tendon unit following massage; the mechanism section below argues that the flexibility effect is better explained neurally.

That last claim can now be stated quantitatively rather than as an absence. Löbell et al. (2026) conducted the first systematic review and meta-analysis dedicated to the acute effects of massage on myotendinous stiffness, pre-registered on PROSPERO and searching five databases plus two preprint servers to 12 January 2026. From 3,584 records they included 25 randomised trials in 617 healthy adults and athletes, covering self-massage (17 studies), manual massage (4) and instrumental techniques (4). Pooling the trials that had a control group or condition (10 studies, 522 participants), the effect on overall muscle stiffness was non-significant (SMD −0.17, 95% CI −0.35 to 0.02; I² = 10%), graded moderate certainty and downgraded only for imprecision. A separate pool for gastrocnemius medialis was likewise null (SMD −0.15, 95% CI −0.58 to 0.29; 5 studies, 230 participants), and subgroup analysis found no difference between instrumental, manual and self-massage. Qualitative synthesis of the remainder found no consistent change in muscle (16 studies), tendon (2) or muscle–tendon unit stiffness (9) — every one of the nine MTU trials was null.

The review also explains why individual trials sometimes report reductions. Among studies finding decreased stiffness, 75% used myotonometers or tissue-hardness meters, which register superficial oscillation or perpendicular displacement and are influenced by skin and subcutaneous tissue; among studies finding no reduction, half used shear-wave elastography, which isolates the shear modulus of deeper muscle. The authors attribute the scattered positive results to assessment instrument, absent controls and differing testing procedures rather than to a consistent biomechanical alteration, and note that most trials never quantified the pressure applied — so the compression thresholds a thixotropic response would require may simply never be reached. Constraints belong on the other side of the ledger too: 83% of outcomes carried "some concerns" on RoB2, only 13 of the 25 trials included a control group at all, and the pooling combined instruments measuring different tissue depths, which the authors flag themselves. Their own clinical inference matches the argument of this document — that increased range of motion, acute analgesia and perceived recovery are more plausibly neural or perceptual than mechanical.

2. Circulatory and "Flushing" Mechanisms

Status: Postulated — largely refuted (the classic "flushes lactate / boosts blood flow" rationale)

The popular idea that massage "flushes out" metabolic waste or increases muscle blood flow has been repeatedly tested and does not hold up. Dakić et al. (2023) concluded that massage did not change blood lactate clearance, muscle blood flow, muscle temperature, or muscle activation. Their underlying tabulation is more mixed than that summary implies, and is worth stating accurately: for lactate removal, 10 studies found no effect against 2 favourable and 1 unfavourable; for blood flow the split was 2 increase, 2 no effect, 1 decrease; and for temperature, studies consistently found a rise in skin temperature with inconsistent effects on deep muscle temperature. The defensible reading is not that every study is null, but that no consistent circulatory effect emerges and that results track the measurement method used. Doppler-ultrasound evidence is the most direct: Wiltshire et al. (2010), measuring the forearm after strenuous isometric handgrip exercise, found that massage impaired lactate and H+ removal from the exercised muscle by mechanically impeding blood flow — the opposite of the intended effect — relative to both passive and active recovery. Reviewing the accumulated blood-flow literature, including repeated Doppler studies of arterial inflow and venous outflow, Tiidus (2026) concluded that muscle blood flow is not influenced by massage and therefore cannot be a factor in any potential recovery benefit. Tiidus (2026) further notes the physiological disconnect that undermines the lactate story independently of massage: lactate clears from muscle within minutes of intense exercise, whereas soreness does not peak until 24–48 h, and eccentric contractions produce substantial soreness with little lactate accumulation. The traditional soigneur rationale — that massage speeds recovery by pumping blood and clearing lactate — is thus contradicted both by direct measurement and by the underlying physiology.

A distinction is needed at this point, because a separate literature reports circulatory effects that are real but not the ones the flushing rationale requires. Bangera, Muralidharan, and Rao (2026), reviewing massage and the circulatory system, describe reductions in blood pressure and heart rate, shifts toward parasympathetic dominance on heart-rate-variability metrics, improved skin blood flow and vascular reactivity after repeated facial roller use, and reduced arterial stiffness with improved flow-mediated dilation in a pilot study — proposing shear-stress-mediated nitric oxide release as the vascular mechanism. None of this contradicts Wiltshire et al. (2010) or Tiidus (2026). Those measured blood flow within the exercised muscle; the effects Bangera et al. assemble are systemic haemodynamic and autonomic changes plus cutaneous perfusion. Massage can plausibly lower blood pressure through a vagal route while failing to raise intramuscular perfusion, and only the latter would support the recovery rationale. The autonomic strand of that review also converges with the parasympathetic evidence in the next section, which is where it does the most work.

The review itself should be weighted accordingly. It is a narrative literature review: no search strategy, eligibility criteria, risk-of-bias assessment or certainty grading is reported, it rests on twelve references, and several of its stronger claims trace to single small studies or pilot data. Its own limitations section is candid about the field — inconsistent description of what "Swedish" or "deep tissue" massage actually involved, small and unrepresentative samples, and the near-impossibility of blinding or placebo control — and it states that the current evidence is insufficient to support routine clinical use for blood pressure. It is cited here for the scope distinction it makes visible, not as an independent estimate of effect.

3. Neurological and Autonomic Mechanisms

Status: Moderate support (pressure-dependent parasympathetic shift; cortisol claim weak)

Here the evidence is more favourable, and it suggests massage acts on the nervous system rather than on muscle tissue per se. Pressure depth appears decisive. Diego and Field (2009) reported that moderate-pressure massage elicited a parasympathetic response — increased high-frequency heart-rate-variability power and a reduced LF/HF ratio, indicating a shift toward vagal dominance — whereas light-pressure massage produced the opposite, a sympathetic response. This provides a plausible physiological basis for the relaxation people report and helps explain why depth of contact matters clinically.

The cortisol mechanism, however, deserves caution because it is widely asserted but weakly supported. Moyer, Seefeldt, Mann, and Jackley (2011), in a comprehensive quantitative review, found massage’s effect on cortisol to be small and generally non-significant — far weaker than the popular "massage lowers cortisol, which explains its benefits" narrative implies. This echoed the earlier meta-analysis of Moyer, Rounds, and Hannum (2004), which found reliable reductions in state anxiety but not in cortisol. Rapaport, Schettler, and Bresee (2012) likewise reported only modest and variable neuroendocrine changes after repeated Swedish massage, in a preliminary dosing study. The autonomic and relaxation pathway is therefore better supported than the specific cortisol-reduction claim built on top of it.

4. Cellular Mechanotransduction and Anti-Inflammatory Mechanisms

Status: Emerging — promising but preliminary (strong mechanistic design, very small sample)

The most mechanistically rigorous human evidence comes from Crane et al. (2012), a within-subject muscle-biopsy study in Science Translational Medicine. In 11 young men who exercised to induce muscle damage, one leg was massaged and the contralateral leg served as an untreated control. Massage activated the mechanotransduction pathways focal adhesion kinase (FAK) and ERK1/2, potentiated mitochondrial-biogenesis signalling (PGC-1α), and blunted the exercise-induced rise in nuclear NF-κB, alongside attenuated production of the inflammatory cytokines IL-6 and TNF-α. Critically, these changes occurred with no effect on muscle metabolites such as glycogen or lactate, reinforcing that any benefit is not metabolic "flushing." This reframes massage as a mechanical signal that modulates inflammatory signalling and supports mitochondrial biogenesis rather than as a circulatory pump.

Three limitations should temper the reading. The sample was 11 men, so precision is low and generalisation to women and to older adults is unwarranted. The comparator was an untreated contralateral limb rather than a sham, so massage-specific effects cannot be separated from any systemic or crossover response. And the outcomes were molecular signals measured at 2.5 h and 24 h post-exercise, not clinical or functional endpoints — the study demonstrates that massage changes intramuscular signalling, not that it accelerates recovery. Consistent with this, the animal and human literature reviewed by Tiidus (2026) shows repeated positive effects of massage-like loading on muscle inflammatory markers in animal models without corresponding gains in the rate of muscle force recovery in humans.

5. Psychological and Perceptual Mechanisms

Status: Well supported (plausibly the dominant driver of benefit)

The most reliable effects of massage are perceptual and psychological — though the sport-specific evidence is thinner than it is usually made to sound. Dakić et al. (2023) reported that massage reduced depression, stress, anxiety, and the perception of fatigue while increasing mood, relaxation, and the perceived quality of recovery. That sentence rests on just 15 studies in total, and the counts behind each individual claim are small: anxiety 4 studies, fatigue perception 7, mood 3 (one of which found no effect), stress 2, perceived recovery 2, and depression, relaxation and positive affect 1 study each. A single uncontrolled study is not a basis for a claim about depression.

The weight here should fall on Moyer, Rounds, and Hannum (2004) instead, whose meta-analysis of 37 RCTs found reductions in trait anxiety and depression to be the largest effects in the massage literature, with reliable single-session reductions in state anxiety. That is a quantitative synthesis of randomised evidence; Dakić et al. corroborates its direction across a broader literature but adds little to its strength. The reasonable conclusion is that anxiety and mood effects are well supported, that perceived recovery is plausibly supported, and that the depression and relaxation claims in the sport literature specifically are not yet established. Several authors argue these perceptual effects are the primary channel through which massage benefits athletes — keeping them relaxed, focused, and subjectively recovered — rather than any direct change in muscle physiology (Dakić et al., 2023; Poppendieck et al., 2016).

Fatigue deserves separating from mood, because the evidence splits by population. In athletes, individual trials report reduced perceived fatigue after massage — including after an Ironman triathlon and a 161 km ultramarathon — but when Davis et al. (2020) pooled the fatigue trials the effect was not significant (SMD 0.47, 95% CI −0.28 to 1.22; 5 studies, n = 171; I² = 86%), and one included study found massage significantly increased perceived fatigue in amateur boxers. In clinical fatigue, by contrast, J. Li et al. (2024) pooled 32 RCTs in 2,594 patients with chronic fatigue syndrome and found substantial benefits on the Fatigue Scale-14: total score MD −1.59 (95% CI −1.84 to −1.34), physical fatigue MD −1.30 (95% CI −1.60 to −1.00), mental fatigue MD −0.85 (95% CI −0.99 to −0.72), and an effective rate of RR 1.23 (95% CI 1.19 to 1.28).

That contrast is striking, and unusually the comparators were active — acupuncture, herbal medicine, moxibustion, psychotherapy, cupping or conventional drugs rather than rest. Four constraints nonetheless limit how far it travels. All 32 trials were conducted in China and published in Chinese-language journals, which the authors identify as a limitation; no trial reported allocation concealment and only one reported any blinding; the FS-14 pools rest on 6 trials (360 participants) for total score and 3 trials (193 participants) each for the mental and physical subscales, with I² of 75–86%; and in 13 of the 32 trials the intervention was massage added to acupuncture versus acupuncture alone, which asks whether massage adds value on top of another therapy rather than whether massage works. The finding is best read as a genuine signal in a distinct clinical population, generated by a literature whose reporting quality the authors themselves describe as insufficiently rigorous — not as evidence that massage relieves exercise-induced fatigue in athletes.

Where the Outcome Evidence Lands

Massage’s best-supported sports outcome is reduced delayed-onset muscle soreness (DOMS), but the size and certainty of that effect depend heavily on which synthesis is read, and the most-cited one has serious methodological problems. Guo et al. (2017) is routinely quoted as showing that massage "alleviates DOMS," reporting pooled effects at 24 h (SMD −0.61), 48 h (SMD −1.51) and 72 h (SMD −1.46) but none immediately post-exercise. Four limitations should temper that reading.

First, the headline sample is inflated. The review reports 23 data points involving 504 participants, yet summing the sample sizes in its own Table 1 gives 239 participants across 11 trials. The 504 figure arises from entering the same individuals at 24, 48 and 72 h as if they were independent — a unit-of-analysis error that artificially narrows confidence intervals and overstates precision.

Second, the comparator is weak. Eight of the eleven trials used no intervention at all (typically seated rest); only three used any form of sham, attention-matched, or active comparator. Expectation and touch effects are therefore largely confounded with any specific effect of massage.

Third, the authors themselves report that the quality of all included trials was low: all used a randomisation method, but none reported allocation concealment, none blinded participants, and only two (18.2%) blinded outcome assessors. Heterogeneity was high (I² = 79% for the pooled soreness outcome; 82% at both 48 and 72 h).

Fourth, the secondary outcomes are fragile. Although the soreness result was stable under leave-one-out sensitivity analysis, the peak-torque finding reversed when a single trial was removed. Guo et al. is therefore best cited as suggestive rather than definitive.

Recent network meta-analysis offers little support for a specific massage effect. Chen et al. (2025) searched eight databases to 22 October 2024 and included 15 RCTs (447 participants) comparing physical-therapy modalities against placebo for DOMS. Only photobiomodulation therapy (24 and 48 h) and sauna (48 h) separated significantly from placebo; beyond 48 h no modality did. Soft-tissue mobilisation — the manual technique closest to massage in that network — showed no significant advantage over placebo at 24 h (−0.89, 95% CrI −2.63 to 0.85). The review did not include manual massage as a labelled comparator, so it should not be cited as direct evidence on massage; it is cited here for the narrower point that manual soft-tissue work does not outperform a credible placebo, and that the apparent DOMS advantage of physical modalities generally shrinks once the comparator is not passive rest.

A second network meta-analysis, published in 2026, puts massage’s soreness effect on a clearer timeline. Hou, Yin, and Qiao (2026) searched five databases to 24 January 2026 and built frequentist networks from 22 randomised trials in team-sport athletes, anchored on passive control and analysed at discrete time points rather than pooled across windows. Massage ranked highest of all modalities for soreness reduction in the acute phase, with standardised mean differences against passive control of −2.24 (95% CI −4.00 to −0.48) immediately post-exercise and −2.62 (95% CI −4.43 to −0.80) at 24 h. By the delayed phase the picture inverts: no intervention retained a significant advantage over control for countermovement jump or creatine kinase, and cold-water immersion was the only modality still associated with lower soreness at 48 h (SMD −1.57, 95% CI −2.58 to −0.56). Massage’s benefit did not persist.

Two features should stop those massage estimates being quoted at face value. Their magnitude sits far outside the rest of the literature: an SMD of −2.6 for soreness is larger than the −1.51 maximum in Guo et al. and more than double the pooled DOMS effect in Davis et al. (1.13). Effects of that size, measured on subjective rating scales in small athlete samples, are more consistent with sparse networks and small-study effects than with an analgesic of that strength. The confidence intervals say as much — the 24 h interval runs from −0.80 to −4.43, so the data are about equally compatible with a modest effect and an implausible one. The authors are candid about why: global heterogeneity exceeded I² = 75% in the soreness and creatine-kinase networks; the sparsity of those networks meant no formal certainty-of-evidence assessment (CINeMA, or GRADE for network meta-analysis) could be computed at all; sensitivity analyses excluding high-risk-of-bias trials and formal publication-bias tests were not feasible; nothing was anchored to a minimal clinically important difference; and the protocol was archived only after data extraction had begun, so it is not a prospective registration. They state explicitly that the ranking scores are not proof of superiority.

What the review does add is a time course and a comparator hierarchy, and both fit the pattern described throughout this document. Massage’s advantage is measured against passive control — the largest body of direct evidence in the network links passive control with cold-water immersion and massage — so this is once again massage beating doing nothing. Where active alternatives compete, massage does not win: active recovery ranked first for restoring explosive performance at 24 h, cold-water immersion first for creatine kinase, and cold-water immersion alone still separated from control at 48 h. The finding that massage’s effect is concentrated in the first 24 h and absent by 48 also sits awkwardly with Guo et al. (2017), whose largest pooled reductions fell at 48 and 72 h; the two syntheses differ in population, comparator structure and time handling too much for the discrepancy to be resolved here, but it is a further reason to treat the size and timing of the DOMS effect as unsettled.

Two further syntheses complete the picture. Dupuy et al. (2018) remains the most useful pragmatic comparison, ranking massage among the more effective options for reducing perceived soreness and fatigue and for lowering inflammatory markers. Poppendieck et al. (2016) found only small performance-recovery effects, most detectable for sprint performance, with modest and variable study quality.

Davis, Alabed, and Chico (2020) is the largest synthesis of manual sports massage to date — 29 randomised studies (12 parallel-group RCTs and 17 randomised crossover trials) recruiting 1,012 participants, with non-manual techniques such as waterjet, foam rolling and automated massage excluded by design. Across seven separately pooled endpoints, five were null: strength (SMD 0.17, 95% CI −0.08 to 0.42; 12 studies, n = 346; I² = 23%), jump (SMD 0.16, 95% CI −0.20 to 0.51; 5 studies, n = 132; I² = 5%), sprint (SMD −0.35, 95% CI −0.98 to 0.28; 7 studies, n = 257; I² = 82%), endurance (SMD 0.21, 95% CI −3.45 to 3.87; 3 studies, n = 96; I² = 97%) and fatigue (SMD 0.47, 95% CI −0.28 to 1.22; 5 studies, n = 171; I² = 86%). Two studies from a single group reported that massage impaired performance — reduced peak isokinetic quadriceps torque after pre-event massage, and reduced electromyographic amplitude after high-intensity exercise — relative to a detuned-ultrasound placebo. The authors conclude that there is no evidence to justify massage where the expectation is direct improvement of performance.

The two positive endpoints require the same scepticism applied to Guo et al. Flexibility (SMD 1.07; 7% improvement) and DOMS (SMD 1.13, 95% CI 0.44 to 1.82; 10 studies, n = 311; 13% improvement) were both statistically significant, but heterogeneity was very high (I² = 90% and 86% respectively) and Davis et al. state explicitly that each pooled estimate was driven by a single outlier study, leaving the true magnitude of benefit uncertain. Two further constraints are stated by the authors themselves: all ten DOMS studies relied on subjective rating scales susceptible to placebo effects, and the flexibility trials compared massage against no intervention rather than against passive stretching or another intervention with its own potential to improve range of motion. A caveat on citation practice is also warranted: the outlier studies are flagged by superscript reference numbers that do not correspond to trials appearing in the relevant forest plots, so the outliers cannot be identified unambiguously from the published text.

Tiidus (2026) reaches the same conclusion from the recovery-physiology side: across systematic reviews and controlled studies, massage has not been shown to enhance the rate of muscle force recovery, the most accurate non-invasive index of post-damage repair, for up to a week following damaging exercise.

The defensible summary is that massage produces a real but modest, delayed reduction in perceived soreness that has not been shown to exceed active recovery, that it does not accelerate the recovery of muscle force, and that the largest reported effects come from the weakest study designs.

Does Massage Work? A Meta-Analytic Verdict by Domain

The sports-recovery meta-analyses above are only part of the picture; massage is used far more widely in clinical care, and the meta-analytic evidence there is best read domain by domain. The recurring pattern is that benefits are real but typically small-to-moderate, strongest for pain and mood, often short-lived, and reported from a literature of generally modest methodological quality and low certainty of evidence. A distinction worth keeping in view is how much of the benefit survives a stringent control: massage most clearly beats no-treatment or passive controls, and its advantage tends to shrink against active or attention-matched comparators — the signature of a benefit that is at least partly non-specific (relaxation, attention, expectation, and touch). The evidence is not uniform on this point; Win Myint et al. (2025) reported that massage improved cancer pain even in comparisons including simple touch and attention controls, which is harder to explain by non-specific effects alone, although every outcome in that review was graded very low certainty.

Musculoskeletal pain

For non-specific low back pain, the Cochrane review by Furlan et al. (2015) concluded that massage may offer short-term pain relief but rated the certainty of evidence as low to very low, with benefits not maintained at longer follow-up and no convincing effect on function. For neck pain, Cheng and Huang (2014) pooled 15 RCTs and found moderate evidence of an immediate benefit for pain compared with inactive therapies (SMD 1.30, 95% CI 0.09 to 2.50), no valid evidence of an effect on dysfunction, and insufficient evidence regarding follow-up effects. Wang, Jiang, and Gao (2022) reached a comparable conclusion for chronic neck pain, finding manual soft-tissue therapy effective for pain in the short term while noting that most included trials were of medium quality. The consistent theme is a genuine but short-lived analgesic effect on pain itself, with weaker evidence for restoring function.

Knee osteoarthritis (KOA) is the musculoskeletal condition for which the newest randomised evidence is available, and it reproduces the pattern closely. Asgarimoghadam, Ravari, Mirzaei, Kamiab, and Abbasifard (2026) randomised 75 adults over 60 years with symptomatic KOA to eight weeks of home-based, self-administered Swedish massage, to hip strengthening exercises, or to a no-intervention control, at three 30-minute sessions per week. Under intention-to-treat ANCOVA adjusted for baseline values, massage outperformed control on all three outcomes: VAS pain −0.81 cm (95% CI −1.24 to −0.38; d = 0.69), KOOS activities-of-daily-living +3.59 points (95% CI +1.62 to +5.56; d = 0.71), and active knee flexion +3.42° (d = 0.73). Hip strengthening produced statistically indistinguishable results, with no significant difference between the two active arms on any outcome (all p > 0.70).

Four features limit how much weight this can bear, and each recurs throughout this document. The comparator was no intervention rather than a sham or attention control, and participants could not be blinded, so expectation effects are confounded with any specific effect — a limitation the authors state first themselves. Both primary outcomes, VAS pain and KOOS-ADL, are self-reported. The absolute effects are small: the 0.81 cm adjusted pain difference falls just below the 0.9-unit difference the investigators nominated as minimally clinically important in their own sample-size calculation, and a 3.6-point gain on a 0–100 KOOS subscale is modest, so the paper’s description of the results as clinically meaningful sits awkwardly with its own threshold. And the massage was self-administered by participants after brief training rather than delivered by a therapist, so the finding does not transfer straightforwardly to clinical practice.

These reservations are not merely inferred here. In a published comment in the same journal, Kumar and Pattnaik (2026) raised three: that the absence of participant blinding alongside subjective outcome measures risks performance bias, which tends to overstate treatment effects in non-pharmacological trials; that the null head-to-head comparison between massage and hip strengthening is underpowered, leaving the equivalence reading vulnerable to Type II error when a formal non-inferiority or equivalence design would have been required to support it; and that screening 710 patients to enrol 75 yields a sample selective enough to constrain external validity. The trial is therefore best read as consistent with the broader picture — a real but small short-term benefit against a passive comparator — rather than as evidence that massage matches an active exercise programme.

Mental health and stress

This is where the effect sizes are largest. The meta-analysis of 37 RCTs by Moyer, Rounds, and Hannum (2004) found that reductions in trait anxiety and depression were massage’s biggest effects, with a course of treatment producing benefits comparable in magnitude to psychotherapy, alongside reliable single-session reductions in state anxiety, blood pressure and heart rate — but no reliable effect on cortisol. That dissociation, a clear psychological benefit without the hypothesised hormonal mechanism, was subsequently confirmed by Moyer et al. (2011) and is one reason the mood effects are best understood as genuine but mechanistically non-specific.

Cancer care and specific conditions

In oncology, massage is used for symptom management rather than as a disease treatment. Win Myint et al. (2025), a systematic review and meta-analysis of 36 RCTs (3,671 participants), found that massage significantly improved cancer pain (pooled SMD −0.51, 95% CI −0.68 to −0.33), quality of life (SMD 0.48) and anxiety (SMD −0.38, 95% CI −0.57 to −0.18) post-intervention. Crucially, every one of these outcomes was graded very low certainty of evidence, with most trials at unclear or high risk of bias and substantial heterogeneity, so the authors explicitly caution that the benefits should be interpreted cautiously. In fibromyalgia, Y.-H. Li et al. (2014) pooled nine RCTs and found that massage sustained for at least five weeks improved pain, anxiety and depression — but not sleep disturbance. Across these conditions the pattern holds: measurable relief of pain and distress, thinner evidence for objective or long-term change, and low certainty throughout.

Efficacy is only half the question, and until recently the harms side of the oncology literature was essentially unexamined. Kjerulf, Christensen, Rørth, Larsen, and Bloomquist (2026) conducted the first systematic review and meta-analysis dedicated to harms of massage in people living with or receiving treatment for cancer, searching ten databases and trial registries to 22 October 2024 under a pre-registered PROSPERO protocol. Their most telling finding concerns the literature rather than the intervention: of 63 intervention studies, 34 (53%) reported nothing at all about adverse events, and among the 29 that did, the median score against the 16-item CONSORT harms extension was 1.5 items. One study reported adverse events by severity and none by seriousness. Where trials in this field report no adverse events, it usually means adverse events were never systematically sought.

Within that constraint the pooled result is reassuring. Across 16 comparisons in 998 participants, the risk ratio for adverse events with massage versus usual care or attention control was 0.69 (95% CI 0.43 to 1.10) with I² = 0%, and the finding held across subgroups defined by cancer stage, cancer treatment, adverse-event severity and risk of bias — though every estimate was graded very low certainty. Two caveats bound it. Massage intensity was light or light-to-moderate in almost all trials, and the three studies describing deep massage reported no adverse-event data, so the review cannot speak to deep-tissue work at all. And among three cohort studies (n = 1,406), two found associations between massage delivered over the area later diagnosed as osteosarcoma and worse survival — both rated at critical risk of bias for confounding, retrospective exposure recall and baseline prognostic imbalance — while a third found no increased breast-cancer recurrence after manual lymphatic drainage (hazard ratio 0.71, 95% CI 0.39 to 1.29). The authors conclude that light-to-moderate massage carries no demonstrated excess risk, that the usual contraindications apply as in the general population, and that massage directly over a tumour should be avoided pending better evidence.

Taken together, the defensible one-line answer is: yes, but modestly and mostly for how people feel. Massage reduces pain and anxiety in the short term and improves perceived recovery and quality of life; its effect on muscle soreness is real but imprecisely estimated; and it does not reliably change strength, muscle function, long-term disability, or the physiological markers (blood flow, lactate, cortisol) once thought to explain it.

Table 1. Meta-analytic verdict by outcome domain

CI = confidence interval; DOMS = delayed-onset muscle soreness; QoL = quality of life; SMD = standardised mean difference. "Supported" denotes consistent meta-analytic benefit on the stated outcome. Most benefits are short-term and of low-to-moderate certainty. Massage typically outperforms no-treatment or passive controls more than it outperforms sham or other active treatments.

How Massage Improves Flexibility — and How It Compares

Flexibility is the one motor outcome massage reliably improves, so it is worth asking precisely how. Whenever range of motion (ROM) increases, there are two candidate mechanisms: a mechanical or structural change (the muscle–tendon unit genuinely lengthens or becomes less stiff) or a sensory and neural change (stretch tolerance — the tissue is unchanged, but a person can move further into range before the stretch sensation or discomfort makes them stop).

The benchmark data come from stretching, the most-studied modality. Ingram et al. (2025), pooling 65 studies and 1,542 adults, found a small reduction in overall stiffness after both acute (Hedges’ g = 0.42) and chronic static stretching (g = 0.37), a moderate increase in stretch tolerance after chronic stretching only (g = 0.74), and no significant effect on fascicle length after either. Improved ROM following chronic stretching was associated with both reduced stiffness and increased stretch tolerance. The authors graded the certainty of evidence as low to very low across all six outcomes. Two points follow for massage: a single session does not lengthen muscle, and tolerance-based mechanisms accumulate over weeks rather than appearing acutely.

The massage-specific mechanism is primarily neural

For massage, the flexibility effect runs mainly through two overlapping neural routes rather than tissue deformation. First, massage acutely dampens the spinal stretch reflex. Behm et al. (2013) showed that massage of the plantar flexors — both at the musculotendinous junction and by tapotement — and static stretching each reduced spinal reflex excitability, indexed by the soleus H-reflex/M-wave ratio, without significantly affecting evoked twitch contractile properties; tapotement produced the greater depression. The muscle’s contractile machinery is unchanged, but the nervous system resists the stretch less — an effect documented for the triceps surae since Morelli, Seaborne, and Sullivan (1991), who recorded reduced soleus H-reflex amplitude during a six-minute manual massage. Weerapong et al. (2005) framed this as the neurological mechanism, whereby pressure alters excitability as indexed by the Hoffmann (H-) reflex, and noted that the direction of the effect depends on technique.

Second, massage produces mild analgesia and a parasympathetic shift (Diego & Field, 2009) that lowers perceived resistance, so more range is tolerated — the same currency stretching trades in. The biomechanical "massage loosens the tissue" account faces a further objection: Beardsley and Škarabot (2015), reviewing self-myofascial release, concluded that the forces required to deform dense connective tissue lie outside the range achievable by manual or roller pressure, which is why the neural and perceptual explanation is favoured. Löbell et al. (2026) supply the direct test that was previously missing: pooled across randomised trials, acute massage does not reduce muscle stiffness (SMD −0.17, moderate certainty), and every trial measuring the muscle–tendon unit was null. Whatever produces the acute range-of-motion gain, it is not a mechanically softer muscle.

Comparison across modalities

The approaches work through substantially overlapping mechanisms. Massage and stretching both reduce reflex excitability while leaving contractile properties intact (Behm et al., 2013); stretching generally yields a larger, more reliable acute ROM gain and is the only modality with meta-analytic evidence of a stiffness change, though even there fascicle length is unaffected (Ingram et al., 2025). A direct comparison is not actually available for massage: Davis et al. (2020) note that the flexibility trials in their meta-analysis compared massage against no intervention rather than against passive stretching, so massage has never been shown to add anything over an alternative that is cheaper and self-administered. A partial exception has since appeared outside the sports literature: Asgarimoghadam et al. (2026) recorded comparable active knee-flexion gains from self-administered massage (+3.42°) and from hip strengthening exercises (+3.69°) in older adults with knee osteoarthritis. That is a genuinely active comparator, but not a flexibility-directed one, and both gains are small, so it narrows the gap without closing it. For foam rolling and self-myofascial release, Konrad et al. (2022) pooled 11 training studies and 46 effect sizes and found that multi-week foam-rolling training increases ROM, with the mechanism attributed to altered pain perception and stretch tolerance rather than to changes in muscle stiffness; Beardsley and Škarabot (2015) had reached the same mechanistic conclusion for acute effects, adding that self-myofascial release increases flexibility and reduces soreness without impeding subsequent performance — its distinctive advantage over prolonged static stretching. Behm and Wilke (2019), reviewing rolling devices specifically, concluded there is insufficient evidence that any literal "release" of myofascia occurs, pointing instead to mechanoreceptor and afferent activation that modulates autonomic tone and engages global pain-modulatory systems, including diffuse noxious inhibitory control. The common thread is that acute flexibility gains are mostly the nervous system permitting more movement, not the tissue becoming longer.

Table 2. Flexibility mechanisms compared across modalities

ROM = range of motion; SMFR = self-myofascial release. Acute effects reflect a single session; structural changes, where present, generally require weeks of repeated training. Evidence for massage and manual myofascial release is weaker and less quantitative than for stretching and foam rolling; the "manual myofascial release" row is extrapolated from the rolling and massage literature rather than from a dedicated meta-analysis.

Muscle Hardening ("Hartspann"): Construct and Evidence

Hartspann (or Muskelhartspann) is a German clinical term for a palpable, often tender hardening or increased tension of a muscle — literally "hard tension." It belongs to a cluster of historical German terms including Myogelose and Muskelhärte, and internationally it maps most closely onto the taut band and the (myofascial) trigger point. In its traditional clinical usage, Hartspann denotes a reflexively mediated increase in muscle tone, distinguished from myogelosis, described as a hardening that persists under anaesthesia. On that reading, Hartspann is a functional tone phenomenon rather than a fixed structural lesion — which is precisely why its proposed response to massage would be neural (reduced reflex tone and autonomic relaxation), consistent with the neurological mechanism described above rather than the biomechanical one.

A caveat is required here. This terminological distinction derives from the German-language clinical tradition and is not, in the sources reviewed for this document, established by peer-reviewed evidence; it is reported as a description of usage, not as a validated construct. The English-language literature discussed below concerns the trigger point and taut band, which are the closest operationalised equivalents.

The core evidence problem: identification and construct validity

The construct rests on manual palpation, whose inter-rater reliability is poor — which is the stated motivation for objective imaging alternatives such as shear-wave elastography (SWE). SWE measurement itself is reproducible in the settings where it has been tested: Shams et al. (2024) reported good-to-excellent intra- and inter-rater reliability for shear modulus of the popliteus and gastrocnemius in 30 participants with knee osteoarthritis accompanied by myofascial trigger points. Two qualifications matter. That study establishes measurement reliability, not that SWE validly identifies trigger points; and a single reliability study in one muscle group and one patient population does not generalise.

The deeper pathophysiological construct is genuinely contested. In an influential critique, Quintner, Bove, and Cohen (2015) argued that the trigger-point and myofascial-pain model lacks external validity and rests partly on circular reasoning — pain is used to infer an "active" point while its absence is attributed to a "latent" one, which makes the model difficult to falsify. Proponents rejected that conclusion, and the debate remains unresolved. The unresolved status is itself the finding: a clinical label that cannot be reliably applied and whose underlying lesion has not been demonstrated is a weak foundation for mechanistic claims about treatment.

What the evidence says about massage’s effect

The picture is modestly positive but low-certainty, and — consistent with the rest of this document — the reliable effects are on pain rather than on demonstrated structural change. Manual soft-tissue therapy for chronic neck pain shows a short-term benefit for pain relief, though most trials are of medium quality (Wang et al., 2022). For trigger-point-directed work specifically, Sadeghnia, Kajbafvala, and Shadmehr (2025) reviewed 12 trials of friction massage and found significant within-group improvements in pain intensity, pressure-pain threshold and range of motion, but no significant superiority over control conditions; the authors graded the short-term evidence for pain and pressure-pain threshold in the upper trapezius as level C and called for well-designed RCTs. The claim that massage reduces objectively measured muscle hardening — as opposed to reducing pain and tenderness — is not supported by the sources reviewed here.

In short: the label describes a real palpable phenomenon, the traditional means of finding it is unreliable, the deeper construct is disputed, and massage’s best-supported effect on it is short-term relief of the associated pain and tenderness rather than a demonstrated resolution of the hardening itself.

The Grand Tour Context: Practice versus Evidence

Elite road cycling is often cited as the paradigmatic case of massage embedded in daily recovery, with riders receiving post-stage massage from team soigneurs throughout a Grand Tour. A caveat must be stated plainly: no peer-reviewed prevalence survey quantifying massage use in professional cycling was identified for this review, and this document therefore makes no quantitative claim about cycling. What can be said is inferential. Given the mechanistic and randomised evidence above, the genuine value of a rider’s daily massage is most plausibly perceptual (reduced soreness, relaxation, and subjective readiness; Dakić et al., 2023; Moyer et al., 2004), possibly involving mechanotransduction-mediated modulation of inflammatory signalling at the tissue level (Crane et al., 2012), and practically useful as a recurring opportunity for staff to examine riders and detect developing injuries. What it is almost certainly not is the traditional flushing-and-blood-flow story, which the circulatory evidence directly contradicts (Wiltshire et al., 2010; Tiidus, 2026).

Why Patients Want and Believe in Massage

The reasons people seek massage only partly overlap with its measurable physiological effects. The demand is sustained by a distinct set of drivers.

Expectation of concrete symptom relief

The most common stated reason is instrumental, and pain relief dominates. In qualitative interviews with 64 patients seeking complementary and alternative treatments for chronic low back pain, Hsu et al. (2014) found that expectations clustered in four domains: pain relief, improved function and the ability to engage in meaningful activities, improved physical fitness, and improved overall well-being. Patients articulated these expectations in terms of desired outcomes rather than mechanisms — a point of some importance, since it means expectation is not contingent on the physiological account being correct.

Touch, comfort, and human connection

This is the theme that most distinguishes massage from other treatments, and much of the qualitative literature identifies it as the real engine of demand. Studying repeat massage users, Smith, Sullivan, and Baxter (2009) found the therapy was valued as a personalised, holistic, hands-on approach centred on relaxation through effective touch within a positive client–therapist relationship and an unhurried environment. They identified six valued elements (time for care and personal attention, an engaging and competent therapist, a trust partnership, holism and empowerment, effective touch, and enhancing relaxation) alongside four modulators — comfort, contact, connection, and caring. The authors argue these humanistic aspects help explain the growing use of massage as much as any physiological effect.

Philosophical congruence and the "pull" toward a holistic model

A large literature on complementary and alternative medicine (CAM) frames belief through a push–pull model: "push" factors repel patients from biomedicine (perceived ineffectiveness, side effects), while "pull" factors draw them toward CAM. In Astin’s (1998) landmark national survey, users of alternative medicine were motivated less by dissatisfaction with conventional care than by finding these approaches more congruent with their own values, beliefs, and philosophical orientation toward health — the finding usually described as philosophical congruence. Zörgő, Peters, and Mkhitaryan (2020) subsequently mapped the specific attitudinal beliefs underlying CAM reliance in 151 participants and found marked differences between biomedically and CAM-oriented groups, supporting the view that reliance tracks a person’s underlying model of health rather than a specific claim about muscle physiology.

Empowerment, expectation, and ritual

Belief is reinforced by a sense of agency — massage gives people something active to do for their own recovery, which matters especially for chronic conditions without a cure. Because massage’s measurable physiological effects are modest, much of the sustained belief plausibly runs through expectancy and the sensory salience of the treatment itself. This is consistent with Moyer et al.’s (2004) finding that massage reliably reduces state anxiety, with Davis et al.’s (2020) caution that soreness and fatigue outcomes are susceptible to placebo effects and biased subjective assessment, and with the sports-massage evidence that its value is largely indirect, keeping athletes relaxed, focused, and subjectively recovered (Dakić et al., 2023).

Massage Across Sports: Prevalence and the Evidence–Practice Gap

Survey data show massage is among the most-used recovery tools across nearly every sport surveyed, though its rated importance varies and consistently outstrips its demonstrated physiological effect. Among 153 elite endurance track-and-field athletes, massage was the second most-used recovery method (86.9%), behind sauna bathing (96.7%) and ahead of daytime napping (81.0%) and long night sleep (61.4%), while better-evidenced methods such as cold-water immersion (15.0%) and compression garments (7.8%) were rarely used; use of massage rose with competitive tier, reaching 100% among elite-tier athletes (Bezuglov et al., 2021). The typical protocol was hand massage delivered by a therapist for 30–60 minutes, at a frequency of either one to two times a week or more than four times a week — a duration substantially longer than the protocols tested in most published trials, which the authors note as a possible source of the discrepancy between evidence and practice.

In a worldwide survey of 107 basketball practitioners, massage was the second most useful-rated strategy (73%, behind active recovery at 80%) but only the fourth most-adopted (61%, behind foam rolling at 72%, active recovery at 68% and stretching at 67%); reported barriers included lack of devices or facilities (51%), excessive cost (51%), lack of time (27%), players’ negative perceptions (25%) and insufficient evidence (16%) (Pernigoni et al., 2022). The authors explicitly frame their findings as a dissociation between established evidence and perceived effectiveness. In rugby, massage was rated more effective by elite than by amateur players, and elite athletes used more recovery modalities more frequently overall (Tavares et al., 2017). In professional football, a worldwide practitioner survey found that 92% of teams used massage — among the most common post-match methods alongside active recovery, cold-water therapy and structured rest (Field et al., 2021).

Two cross-cutting patterns emerge. First, a consistent evidence–practice gap: massage’s popularity exceeds its demonstrated physiological effect in every sport surveyed, echoing Poppendieck et al.’s (2016) observation that an athlete simply feeling better after massage may be sufficient to justify its use. Second, usage scales with resources and status — penetration is highest where teams fund therapists and where athletes compete at higher tiers (Bezuglov et al., 2021; Pernigoni et al., 2022), and cost and facilities are the most frequently cited barriers.

Table 3. Reported massage use across sports (peer-reviewed surveys)

Percentages are not directly comparable across studies: some report the proportion of individual athletes using a method, others the proportion of teams or of surveyed practitioners adopting it. Professional road cycling is deliberately omitted — no peer-reviewed prevalence survey quantifying massage frequency in elite cycling was identified for this review, and narrative descriptions of near-universal daily use are not survey-derived.

Summary: Evidence Strength by Mechanism

DOMS = delayed-onset muscle soreness; FAK = focal adhesion kinase; NF-κB = nuclear factor kappa B; PGC-1α = peroxisome proliferator-activated receptor gamma coactivator 1-alpha.

Overall conclusion. Massage has a defensible evidence base for what people feel and a thin one for what happens in the muscle. Its most secure effects — reduced anxiety, improved mood and perceived recovery, short-term analgesia — are mediated most plausibly by neural and psychological pathways, not by circulation, lactate clearance, cortisol, or tissue deformation — the last of which is now a meta-analytic null rather than merely undemonstrated. The two performance-adjacent outcomes that do reach significance, flexibility and DOMS, are each outlier-driven with I² around 90%, measured on subjective scales in the case of DOMS, and established only against no-treatment comparators. Every measure of actual performance — strength, jump, sprint, endurance — is null, and two trials point the other way. The mechanotransduction evidence is the most interesting open question, but it rests on molecular endpoints in eleven participants and has not been linked to functional recovery. The gap between how much massage is used and how much it has been shown to do is large, consistent, and unlikely to be closed by the existing trial designs, most of which compare massage against doing nothing.

References

All 43 sources are peer-reviewed journal articles. They are of mixed design — meta-analyses and systematic reviews, individual randomised trials, narrative reviews, and one correspondence item — and no uniform standard of evidence should be inferred from inclusion in this list. The design, comparator and limitations of each source are stated at the point where it is cited. Each DOI was checked against the publisher record, and volume, issue, page ranges and author lists were verified against the article of record; the sole exception is Löbell et al. (2026), whose publisher blocks automated resolution and which was verified against the article of record and the journal contents listing. No books, book chapters, websites, news sources, or non-peer-reviewed material are cited. The links in this reference list resolve through the DOI service to the publishers' pages. Some of those pages are hosted outside Switzerland and the EU. Following a link transmits your IP address to the provider concerned — nothing of the kind happens on our own page.

Asgarimoghadam, A., Ravari, A., Mirzaei, T., Kamiab, Z., & Abbasifard, M. (2026). Swedish massage versus hip strengthening exercises for pain and function in older adults with knee osteoarthritis: A randomized controlled trial. Aging Clinical and Experimental Research, 38(1), Article 42. https://doi.org/10.1007/s40520-025-03313-z

Astin, J. A. (1998). Why patients use alternative medicine: Results of a national study. JAMA, 279(19), 1548–1553. https://doi.org/10.1001/jama.279.19.1548

Bangera, D., Muralidharan, S., & Rao, R. M. (2026). Effect of massage therapy on the circulatory system: A literature review. International Journal of Therapeutic Massage & Bodywork, 19(2), 87–91. https://doi.org/10.3822/ijtmb.v19i2.1343

Beardsley, C., & Škarabot, J. (2015). Effects of self-myofascial release: A systematic review. Journal of Bodywork and Movement Therapies, 19(4), 747–758. https://doi.org/10.1016/j.jbmt.2015.08.007

Behm, D. G., Peach, A., Maddigan, M., Aboodarda, S. J., DiSanto, M. C., Button, D. C., & Maffiuletti, N. A. (2013). Massage and stretching reduce spinal reflex excitability without affecting twitch contractile properties. Journal of Electromyography and Kinesiology, 23(5), 1215–1221. https://doi.org/10.1016/j.jelekin.2013.05.002

Behm, D. G., & Wilke, J. (2019). Do self-myofascial release devices release myofascia? Rolling mechanisms: A narrative review. Sports Medicine, 49(8), 1173–1181. https://doi.org/10.1007/s40279-019-01149-y

Bezuglov, E., Lazarev, A., Khaitin, V., Chegin, S., Tikhonova, A., Talibov, O., Gerasimuk, D., & Waśkiewicz, Z. (2021). The prevalence of use of various post-exercise recovery methods after training among elite endurance athletes. International Journal of Environmental Research and Public Health, 18(21), Article 11698. https://doi.org/10.3390/ijerph182111698

Chen, J., Hu, Q., Hu, J., Liu, S., & Yin, L. (2025). Differences in the effectiveness of different physical therapy modalities in the treatment of delayed-onset muscle soreness: A systematic review and Bayesian network meta-analysis. Journal of Pain Research, 18, 2993–3008. https://doi.org/10.2147/JPR.S519242

Cheng, Y. H., & Huang, G. C. (2014). Efficacy of massage therapy on pain and dysfunction in patients with neck pain: A systematic review and meta-analysis. Evidence-Based Complementary and Alternative Medicine, 2014, Article 204360. https://doi.org/10.1155/2014/204360

Crane, J. D., Ogborn, D. I., Cupido, C., Melov, S., Hubbard, A., Bourgeois, J. M., & Tarnopolsky, M. A. (2012). Massage therapy attenuates inflammatory signaling after exercise-induced muscle damage. Science Translational Medicine, 4(119), 119ra13. https://doi.org/10.1126/scitranslmed.3002882

Dakić, M., Toskić, L., Ilić, V., Đurić, S., Dopsaj, M., & Šimenko, J. (2023). The effects of massage therapy on sport and exercise performance: A systematic review. Sports, 11(6), Article 110. https://doi.org/10.3390/sports11060110

Davis, H. L., Alabed, S., & Chico, T. J. A. (2020). Effect of sports massage on performance and recovery: A systematic review and meta-analysis. BMJ Open Sport & Exercise Medicine, 6(1), Article e000614. https://doi.org/10.1136/bmjsem-2019-000614

Diego, M. A., & Field, T. (2009). Moderate pressure massage elicits a parasympathetic nervous system response. International Journal of Neuroscience, 119(5), 630–638. https://doi.org/10.1080/00207450802329605

Dupuy, O., Douzi, W., Theurot, D., Bosquet, L., & Dugué, B. (2018). An evidence-based approach for choosing post-exercise recovery techniques to reduce markers of muscle damage, soreness, fatigue, and inflammation: A systematic review with meta-analysis. Frontiers in Physiology, 9, Article 403. https://doi.org/10.3389/fphys.2018.00403

Field, A., Harper, L. D., Chrismas, B. C. R., Fowler, P. M., McCall, A., Paul, D. J., Chamari, K., & Taylor, L. (2021). The use of recovery strategies in professional soccer: A worldwide survey. International Journal of Sports Physiology and Performance, 16(12), 1804–1815. https://doi.org/10.1123/ijspp.2020-0799

Furlan, A. D., Giraldo, M., Baskwill, A., Irvin, E., & Imamura, M. (2015). Massage for low-back pain. Cochrane Database of Systematic Reviews, 2015(9), Article CD001929. https://doi.org/10.1002/14651858.CD001929.pub3

Guo, J., Li, L., Gong, Y., Zhu, R., Xu, J., Zou, J., & Chen, X. (2017). Massage alleviates delayed onset muscle soreness after strenuous exercise: A systematic review and meta-analysis. Frontiers in Physiology, 8, Article 747. https://doi.org/10.3389/fphys.2017.00747

Hou, C., Yin, W., & Qiao, F. (2026). Acute and delayed effects of post-exercise recovery strategies on explosive performance and markers of muscle damage: A systematic review and network meta-analysis. Healthcare, 14(10), Article 1321. https://doi.org/10.3390/healthcare14101321

Hsu, C., Sherman, K. J., Eaves, E. R., Turner, J. A., Cherkin, D. C., Cromp, D., Schafer, L., & Ritenbaugh, C. (2014). New perspectives on patient expectations of treatment outcomes: Results from qualitative interviews with patients seeking complementary and alternative medicine treatments for chronic low back pain. BMC Complementary and Alternative Medicine, 14, Article 276. https://doi.org/10.1186/1472-6882-14-276

Ingram, L. A., Tomkinson, G. R., d’Unienville, N. M. A., Gower, B., Gleadhill, S., Boyle, T., & Bennett, H. (2025). Mechanisms underlying range of motion improvements following acute and chronic static stretching: A systematic review, meta-analysis and multivariate meta-regression. Sports Medicine, 55(6), 1449–1466. https://doi.org/10.1007/s40279-025-02204-7

Kjerulf, N., Christensen, J., Rørth, M., Larsen, A., & Bloomquist, K. (2026). Harms of massage in persons with cancer or receiving treatment for cancer: A systematic review and meta-analysis. PLOS ONE, 21(8), Article e0351878. https://doi.org/10.1371/journal.pone.0351878

Konrad, A., Nakamura, M., Tilp, M., Donti, O., & Behm, D. G. (2022). Foam rolling training effects on range of motion: A systematic review and meta-analysis. Sports Medicine, 52(10), 2523–2535. https://doi.org/10.1007/s40279-022-01699-8

Kumar, S., & Pattnaik, S. (2026). Comment on "Swedish massage versus hip strengthening exercises for pain and function in older adults with knee osteoarthritis: A randomized controlled trial" [Letter]. Aging Clinical and Experimental Research, 38(1), Article 167. https://doi.org/10.1007/s40520-026-03418-z

Li, J., Piao, F., Zeng, Q., Yan, H., Bi, Y., Zhang, S., & Song, B. (2024). The effect of massage on patients with chronic fatigue syndrome: A systematic review and meta-analysis. Medicine, 103(18), Article e37973. https://doi.org/10.1097/MD.0000000000037973

Li, Y.-H., Wang, F.-Y., Feng, C.-Q., Yang, X.-F., & Sun, Y.-H. (2014). Massage therapy for fibromyalgia: A systematic review and meta-analysis of randomized controlled trials. PLOS ONE, 9(2), Article e89304. https://doi.org/10.1371/journal.pone.0089304

Löbell, R., Glänzel, M. H., Prates, I. M., Petter, G. do N., Marqueti, R. de C., de Oliveira, R. R., & Geremia, J. M. (2026). Short-term effectiveness of massage techniques on myotendinous structures’ stiffness: A systematic review and meta-analysis. Research in Sports Medicine. Advance online publication. https://doi.org/10.1080/15438627.2026.2697196

Morelli, M., Seaborne, D. E., & Sullivan, S. J. (1991). H-reflex modulation during manual muscle massage of human triceps surae. Archives of Physical Medicine and Rehabilitation, 72(11), 915–919. https://doi.org/10.1016/0003-9993(91)90011-7

Moyer, C. A., Rounds, J., & Hannum, J. W. (2004). A meta-analysis of massage therapy research. Psychological Bulletin, 130(1), 3–18. https://doi.org/10.1037/0033-2909.130.1.3

Moyer, C. A., Seefeldt, L., Mann, E. S., & Jackley, L. M. (2011). Does massage therapy reduce cortisol? A comprehensive quantitative review. Journal of Bodywork and Movement Therapies, 15(1), 3–14. https://doi.org/10.1016/j.jbmt.2010.06.001

Pernigoni, M., Conte, D., Calleja-González, J., Boccia, G., Romagnoli, M., & Ferioli, D. (2022). The application of recovery strategies in basketball: A worldwide survey. Frontiers in Physiology, 13, Article 887507. https://doi.org/10.3389/fphys.2022.887507

Poppendieck, W., Wegmann, M., Ferrauti, A., Kellmann, M., Pfeiffer, M., & Meyer, T. (2016). Massage and performance recovery: A meta-analytical review. Sports Medicine, 46(2), 183–204. https://doi.org/10.1007/s40279-015-0420-x

Quintner, J. L., Bove, G. M., & Cohen, M. L. (2015). A critical evaluation of the trigger point phenomenon. Rheumatology, 54(3), 392–399. https://doi.org/10.1093/rheumatology/keu471

Rapaport, M. H., Schettler, P., & Bresee, C. (2012). A preliminary study of the effects of repeated massage on hypothalamic–pituitary–adrenal and immune function in healthy individuals: A study of mechanisms of action and dosage. Journal of Alternative and Complementary Medicine, 18(8), 789–797. https://doi.org/10.1089/acm.2011.0071

Sadeghnia, M., Kajbafvala, M., & Shadmehr, A. (2025). The effect of friction massage on pain intensity, PPT, and ROM in individuals with myofascial trigger points: A systematic review. BMC Musculoskeletal Disorders, 26(1), Article 250. https://doi.org/10.1186/s12891-025-08372-x

Shams, M., Karimi, N., Vahedi, M., Kamali Hakim, P., Zeinalkhani, F., & Rahnama, L. (2024). Reliability of muscle stiffness measures in popliteus, medial and lateral gastrocnemius muscles by ultrasound shear wave elastography in participants with knee osteoarthritis accompanied by myofascial trigger points. BMC Musculoskeletal Disorders, 25(1), Article 221. https://doi.org/10.1186/s12891-024-07351-y

Smith, J. M., Sullivan, S. J., & Baxter, G. D. (2009). The culture of massage therapy: Valued elements and the role of comfort, contact, connection and caring. Complementary Therapies in Medicine, 17(4), 181–189. https://doi.org/10.1016/j.ctim.2009.05.003

Tavares, F., Healey, P., Smith, T. B., & Driller, M. (2017). The usage and perceived effectiveness of different recovery modalities in amateur and elite rugby athletes. Performance Enhancement & Health, 5(4), 142–146. https://doi.org/10.1016/j.peh.2017.04.002

Tiidus, P. M. (2026). Mechanisms and efficacy of massage therapy for post-exercise muscle repair: A narrative review. Muscles, 5(2), Article 29. https://doi.org/10.3390/muscles5020029

Wang, S.-Q., Jiang, A.-Y., & Gao, Q. (2022). Effect of manual soft tissue therapy on the pain in patients with chronic neck pain: A systematic review and meta-analysis. Complementary Therapies in Clinical Practice, 49, Article 101619. https://doi.org/10.1016/j.ctcp.2022.101619

Weerapong, P., Hume, P. A., & Kolt, G. S. (2005). The mechanisms of massage and effects on performance, muscle recovery and injury prevention. Sports Medicine, 35(3), 235–256. https://doi.org/10.2165/00007256-200535030-00004

Win Myint, O., Yoong, S. Q., Toh, E., Lei, F., & Jiang, Y. (2025). Effectiveness of massage therapy for cancer pain, quality of life and anxiety levels: A systematic review and meta-analysis. Journal of Clinical Nursing, 34(1), 49–87. https://doi.org/10.1111/jocn.17547

Wiltshire, E. V., Poitras, V., Pak, M., Hong, T., Rayner, J., & Tschakovsky, M. E. (2010). Massage impairs postexercise muscle blood flow and "lactic acid" removal. Medicine & Science in Sports & Exercise, 42(6), 1062–1071. https://doi.org/10.1249/MSS.0b013e3181c9214f

Zörgő, S., Peters, G.-J. Y., & Mkhitaryan, S. (2020). Attitudes underlying reliance on complementary and alternative medicine. Integrative Cancer Therapies, 19, Article 1534735420910472. https://doi.org/10.1177/1534735420910472