Lactate threshold analyser

Evidence-based tool for analysing lactate step tests and calibrating training zones for cycling (e.g. Rouvy).

🚧 Beta version: This tool is under active development. Methods, thresholds and content are continuously reviewed and updated.

✉ Send feedback

🔒 Privacy: This tool does not store any data. All data you enter or upload are processed exclusively in your browser. Nothing is sent to or stored on any server. When you close or reload the page, all data are permanently gone. If you would like to keep your data, download the Excel template, enter your measurements there, and save the file on your computer – you can upload it at any time to view your results again.

How to use this tool

① Direct data entry

Scroll down to "Step data" and enter power (W) and lactate (mmol/L) for each stage – both are required fields. Heart rate and RPE are optional but unlock extra charts. Then click "Analyse".

→ For a single test, quick and easy

② Excel upload (1 test)

Download the Excel template (button below). Fill in your data in the "Test_1" tab – athlete name, date and measurements. Upload the file under "Upload single test". The tool loads and analyses the test automatically – just like direct entry.

→ For structured data entry outside the browser

③ Excel upload (multiple tests)

Fill the template with multiple tabs (Test_1, Test_2, Test_3…), one tab per test and athlete. Upload under "Upload multiple tests". All tests are then available for comparison and longitudinal analysis.

→ For cross-sectional and longitudinal athlete analysis

④ Sample data

Three pre-built athlete profiles from published literature – a recreational cyclist, a well-trained cyclist, and an elite cyclist. The well-trained profile includes two tests illustrating a rightward curve shift after polarised training (Neal et al., 2013). Click "Analyse" on any single profile, or "Load all" for the full comparison and longitudinal view.

→ To explore the tool without your own data

⑤ After analysis

Below the threshold cards, use the LT1 method selector to choose which LT1 methods appear as lines in all charts — up to 3 simultaneously, each in a distinct colour. The first selected method anchors your training zones and recommendations. Then pick a zone model (Coggan, Seiler, Norwegian) and enter your estimated TTE to calibrate FTP.

→ Refine which LT1 estimate best fits your physiology

What this tool offers

📈 Threshold analysis
Automatically computes 16 lactate threshold methods across 6 families: Bsln+ (0.3–1.5 mmol/L), OBLA (2.0–4.0 mmol/L), D-max variants (nadir, classic, modified), Log-log (LT1 & LT2), LTP (segmented regression), and LTratio. All methods validated against the lactater R package reference dataset. Select 1–3 LT1 methods to display simultaneously as chart lines; the first selected method anchors your training zones.

🏋️ Zone models
Four zone systems to choose from: Rouvy/Coggan 7-zone, Seiler 3-zone, Coggan 5-zone, and the Norwegian 5-zone model – each anchored to your individual LT1 and LT2.

⚡ FTP calibration
Enter how long you could sustain your LT2 power (TTE in minutes) to compute a corrected FTP using the power-duration relationship (Coggan & Allen, 2010).

📄 Training recommendations
Evidence-based training advice tailored to your thresholds and chosen zone model – downloadable as a PDF for use with your athletes or for your own training.

📊 Charts & export
Lactate curve, log-log curve, heart rate, RPE, lactate economy, Δ-lactate, and cardiac efficiency – each showing your selected LT1 method lines and LT2. All charts exportable as PNG. Full results (all 16 methods) exportable as Excel. References downloadable as .ris for Zotero, Mendeley, or EndNote.

🔬 Scientific background
Detailed explanations of training models (polarised, Norwegian, threshold), the physiology of fat oxidation, FatMax, the Talk Test, and over 50 peer-reviewed references.

📥 Download Excel template

The template includes an instructions tab, a filled example (Test_1) and an empty second tab (Test_2). Columns: Step · Power (W) · Lactate (mmol/L) · HR (bpm) · RPE (Borg). Duplicate tabs for additional tests.

② Upload single test & analyse

📊

Drop Excel here (1 test)

Analysed immediately like direct entry

③ Upload multiple tests (comparison)

📂

Drop Excel here (multiple tests)

Loads all tabs for comparison & trends

④ Sample data

Note: These curves are constructed from published group means as representative example profiles – not raw data from individual athletes. References are shown on each card.

Test comparison & longitudinal trends

Compare two tests

● Test A

vs.

● Test B

Longitudinal trend

Select an athlete to display LT1/LT2 development across all loaded tests over time.

Step data

Step	Power (W) *	Lactate (mmol/L) *	HR (bpm)	RPE (0–10)

Shift from row:

Training models — polarized & Norwegian

Scientific background — lactate, FatMax, Talk Test & evidence

Threshold estimates

LT1 — Aerobic threshold methods

Bsln +0.5 mmol/L (primary)

LT1 is the power at which blood lactate rises 0.5 mmol/L above a resting baseline. Two baseline anchors are available — choose based on your protocol:
• Resting (step₁, poly): Baseline = lactate at the first measured step; threshold crossed on the polynomial curve with sub-step interpolation. Matches the original Berg et al. (1990) definition and the lactater R package. Best when your first step genuinely represents resting or near-resting lactate.
• Nadir (step-quantised): Baseline = the lowest observed lactate, wherever it falls in the test. LT1 = first measured step at or above nadir + 0.5 mmol/L. More appropriate when the curve shows a U-shape (lactate dips at low intensities before rising), because anchoring to data[0] would overstate the true minimum baseline.

Rationale: Meixner et al. (2025) validated Bsln+0.5 as the most reliable individual Zone 2 ceiling marker in 50 cyclists (CV 6–8%, Bland–Altman bias −0.7 W vs VT1). The baseline anchor choice can shift results by 0–20 W depending on curve shape — the selector exposes both so you can compare directly on your data.

Berg, A. et al. (1990). PMID 2408033 · Meixner, B. et al. (2025). Transl Sports Med, 2008291. [DOI] · Jones AM & Doust JH (1998). Med Sci Sports Exerc 30(8):1304.

Bsln +0.3 mmol/L

A more conservative threshold requiring only a 0.3 mmol/L rise above baseline. The same two anchor choices apply:
• Nadir (step-quantised): Default for this delta — anchors to the minimum observed lactate and returns the first measured step at or above threshold. Robust against polynomial edge artefacts that can occur when delta is very small.
• Resting (step₁, poly): Polynomial fit with data[0] as baseline — matches lactater's approach and recovers sub-step precision, but requires a sanity check (result must be after the first step) to filter artefacts on some curve shapes.

Rationale: Recommended only with high-precision analysers (e.g. YSI 2300, EKF Biosen) where noise is below 0.3 mmol/L. Capillary point-of-care methods (Lactate Pro, Accutrend) have typical analytical error of ±0.2–0.5 mmol/L, making a 0.3 mmol/L rise potentially indistinguishable from noise.

Faude, O., Kindermann, W., & Meyer, T. (2009). Sports Medicine, 39(6), 469–490. [DOI]

Bsln +1.0 / +1.5 mmol/L

Larger fixed delta above the individual minimum baseline. A 1.0 or 1.5 mmol/L rise requires more accumulation before LT1 is identified, placing the threshold later in the test — often closer to what some protocols call an "individual anaerobic threshold" or a moderately elevated but still sustainable intensity.

Rationale: Zoladz et al. proposed these larger deltas to align more closely with ventilatory threshold in trained athletes, where the lactate curve rises slowly and a 0.5 mmol/L rise can occur well below true physiological threshold intensity.

Zoladz, J.A. et al. (1995). Journal of Physiology, 489(3), 905 · Faude et al. (2009). Sports Medicine, 39(6), 469.

Log-log (Beaver 1985)

The lactate–power curve is log-transformed on both axes (ln[power] vs ln[lactate]) and fitted with a polynomial. LT1 is the power at the point of maximum perpendicular distance from the line connecting the first and last log-log data points (geometric D-max). The search is restricted to values after the curve nadir, ensuring LT1 falls on the rising portion of the curve.

Rationale: Logarithmic transformation linearises the characteristic J-shaped lactate curve, making the first inflection geometrically detectable. Meixner et al. (2025) reported the highest correlation between log-log LT1 and FatMax (r = 0.65) among all LT1 methods tested. Note: lactater uses segmented regression rather than geometric D-max in log-log space, explaining the algorithm divergence seen in validation.

Beaver, W.L., Wasserman, K., & Whipp, B.J. (1985). J Appl Physiol, 59(6), 1936–1940. [DOI]

LTP1 — Lactate Turn Point 1 (Hofmann & Tschakert 2017)

A segmented (hinge) regression with two continuous breakpoints is fitted to the polynomial lactate curve. LTP1 is the first breakpoint — the power at which the lactate–power relationship changes slope for the first time. The two-knot model is optimised by exhaustive search over all valid knot pairs, minimising the residual sum of squares.

Rationale: Breakpoint detection is model-driven rather than threshold-dependent, making LTP1 independent of resting lactate levels. Important caveat: lactater anchors segmented regression to a resting (0 W) step not present in most field protocols — this shifts LTP1 leftward in lactater relative to this tool's implementation, explaining the larger divergence seen in validation.

Hofmann, P. & Tschakert, G. (2017). Front Physiol, 8, 337. [DOI]

LTratio — Lactate/Power ratio minimum (Dickhuth 1999)

The lactate–power ratio (mmol/L per W) is computed at each step. LT1 is the power at which this ratio reaches its minimum on the fitted polynomial — the nadir of the lactate economy curve. After this point, lactate rises faster than power output, indicating increasing glycolytic contribution.

Rationale: The minimum lactate/power ratio represents the most aerobically efficient intensity before glycolytic metabolism begins to dominate. Dickhuth et al. called this the "individual aerobic threshold." This method is more sensitive to curve shape than to absolute lactate levels. Note: lactater uses B-spline fitting for LTratio; this tool uses polynomial fitting, which can give modestly different results.

Dickhuth, H.H. et al. (1999). J Sports Med Phys Fitness, 39(1), 9–14.

LT2 — Anaerobic threshold / MLSS methods

All LT2 methods target the same physiological boundary — the maximal lactate steady state (MLSS) — but differ in algorithm and sensitivity. Valenzuela et al. (2021) showed that D-max LT2 correlates closely with Critical Power at the group level (mean bias ~0.5 W) but with wide individual limits of agreement (±52 W). Zone prescriptions from any LT2 method should be treated as a starting point and refined with perceived exertion and training response. [DOI]

D-max (nadir) (primary)

A polynomial (degree ≤ 4) is fitted to the lactate curve. The curve minimum (nadir) on the polynomial serves as the left anchor; the last measured point is the right anchor. LT2 is the power at maximum perpendicular distance from the straight line connecting these two points. This is the default LT2 method in this tool.

Rationale: Using the polynomial nadir rather than the first data point as the left anchor makes this variant more robust when early lactate measurements are elevated from warm-up effects or incomplete rest. In the lactater validation dataset, this implementation matches the lactater reference to within 0–1 W.

Cheng, B. et al. (1992). Int J Sports Med, 13(7), 518–522. [DOI]

D-max (classic, Cheng 1992)

The original Cheng et al. method: the line runs from the first to the last measured data point (not the polynomial nadir). LT2 is the power of maximum perpendicular distance from this line on the fitted polynomial. The most widely cited and validated D-max variant.

Rationale: Anchoring to the first measured point (rather than the nadir) makes classic D-max sensitive to elevated starting lactate from insufficient warm-up — the anchor point is too high, pulling the line upward and shifting LT2 leftward. Best used when the first step genuinely represents resting lactate. Matches lactater to within 0–1 W in validation.

Cheng, B. et al. (1992). Int J Sports Med, 13(7), 518–522. [DOI]

D-max modified (Bishop 1998)

A modification of classic D-max where the left anchor is the last data point before the first step showing a lactate rise of >0.4 mmol/L above the minimum — rather than the absolute first point. This makes the left anchor more physiologically meaningful and less sensitive to initial fluctuations.

Rationale: Bishop et al. proposed this modification specifically to reduce the influence of the arbitrary choice of the first workload on D-max placement. Tends to give higher (rightward) LT2 estimates than classic D-max on datasets with gradually rising early lactate.

Bishop, D., Jenkins, D.G., & Mackinnon, L.T. (1998). Med Sci Sports Exerc, 30(8), 1270–1275. [DOI]

Log-Poly-ModDmax (Jamnick 2018)

The log-log polynomial LT1 is used as the left anchor; D-max is applied from this point to the last measured point on the original (non-log) polynomial curve. Combines the log-log transformation's sensitivity for early-curve inflection with the D-max geometry for the LT2 region.

Rationale: Jamnick et al. showed this combination reduces bias relative to Critical Power compared to classic D-max variants in trained cyclists. It tends to produce higher LT2 estimates than D-max classic/nadir, placing the threshold closer to the steep rise of the lactate curve.

Jamnick, N.A. et al. (2018). PLOS ONE, 13(7), e0199794. [DOI]

Log-log LT2 (Beaver 1985)

In log-log space, the line runs from the log-log LT1 point to the last measured point. LT2 is the power of maximum perpendicular distance from this line on the log-log polynomial. A geometric cross-validation method that operates entirely in transformed coordinates. No direct equivalent in the lactater R package.

Rationale: Beaver et al. described using the second inflection in log-log transformed gas exchange data to identify the ventilatory threshold — this tool applies the analogous geometric D-max approach to log-log lactate data. Consistently gives higher LT2 estimates (~8–10 W) than D-max classic, within the known inter-method limits of agreement of ±52 W (Valenzuela et al. 2021).

Beaver, W.L. et al. (1985). J Appl Physiol, 59(6), 1936–1940. [DOI]

LTP2 — Lactate Turn Point 2 (Hofmann & Tschakert 2017)

The second breakpoint in the segmented two-breakpoint regression model (see LTP1 above). LTP2 corresponds to the power at which the lactate–power slope changes for the second time — the onset of the steep, non-linear lactate rise above the MLSS boundary.

Rationale: The two-breakpoint hinge model directly detects the transition from moderate to steep lactate accumulation without requiring a fixed concentration target. LTP2 tends to give higher estimates than D-max methods and is considered by Hofmann & Tschakert to correspond more closely to the anaerobic threshold than to MLSS per se.

Hofmann, P. & Tschakert, G. (2017). Front Physiol, 8, 337. [DOI]

OBLA — Onset of Blood Lactate Accumulation

OBLA methods identify the power at which the polynomial lactate curve crosses a fixed absolute concentration. Unlike anchored methods (Bsln+, D-max), OBLA thresholds do not adapt to individual resting lactate and therefore show higher inter-individual variability (Meixner et al. 2025). They are most useful for between-athlete comparisons and longitudinal tracking when resting lactate is stable.

OBLA 2.0 mmol/L

Power at the 2.0 mmol/L threshold on the polynomial fit. Proposed by Heck et al. as a lower-bound aerobic marker, sometimes used to define the upper boundary of Zone 1 in threshold-based models. In well-trained athletes with low resting lactate, this may underestimate the true MLSS.

Heck, H. et al. (1985). Int J Sports Med, 6(3), 117–123. [DOI]

OBLA 2.5 mmol/L

Power at the 2.5 mmol/L threshold. Used in several European sports medicine protocols as the aerobic threshold marker, often considered a reasonable approximation of LT1 in moderately trained individuals. Kindermann et al. originally proposed values in this range as the "aerobic threshold."

Kindermann, W. et al. (1979). Eur J Appl Physiol, 42, 25–34. [DOI]

OBLA 3.0 mmol/L

Power at 3.0 mmol/L. An intermediate marker sometimes used to approximate MLSS in recreational athletes. Population studies have shown 3.0 mmol/L to correlate well with MLSS in heterogeneous cohorts, though individual variability remains high.

Kindermann, W. et al. (1979). Eur J Appl Physiol, 42, 25–34.

OBLA 3.5 mmol/L

Power at 3.5 mmol/L. Closer to the classic 4.0 mmol/L anaerobic threshold. May better estimate MLSS in well-trained athletes who can sustain relatively high lactate concentrations at steady state.

Faude, O., Kindermann, W., & Meyer, T. (2009). Sports Med, 39(6), 469–490.

OBLA 4.0 mmol/L ("anaerobic threshold")

The classic fixed-concentration anaerobic threshold, popularised by Heck et al. (1985) and Mader et al. Long used as the primary LT2 marker in European sports medicine. Despite widespread use, Faude et al. (2009) reviewed evidence that 4.0 mmol/L overestimates MLSS in elite athletes (who can sustain lactate above 4.0 at steady state) and underestimates it in sedentary individuals.

Heck, H. et al. (1985). Int J Sports Med, 6(3), 117–123. [DOI] · Faude et al. (2009). Sports Med, 39(6), 469.

Charts

Why LT1 anchors the Z2 ceiling — and not 75% of FTP (Meixner et al., 2025). A controlled laboratory study of 50 cyclists compared nine common Zone 2 markers (HRmax percentages, fixed blood lactate concentrations, VT1, FatMax, etc.) for interindividual variability. Coefficients of variation for power output ranged from 6% to 29% — fixed %HRmax targets (HR72%, HR82%) and fixed lactate concentrations (e.g. 2 mmol/L) showed the greatest individual scatter. VT1 (≈ LT1) and BLamin+0.5 (baseline + 0.5 mmol/L) showed the closest mutual agreement (Bland-Altman bias −0.7 W, LoA ±37 W) and are considered the most physiologically valid markers for the Z2 upper boundary.
This tool anchors Z2 directly to your measured LT1 — rather than the population-average 75% of FTP. This is not just a methodological choice; it is supported by current evidence.

FTP, Critical Power and MLSS — clarifying the concepts. FTP (Functional Threshold Power) and Critical Power (CP) target the same physiological boundary — the maximal lactate steady state (MLSS) — but are not identical. CP, derived from exhaustive time trials using the power-duration hyperbolic model, typically sits slightly above FTP (~5–10 W in trained cyclists; Karsten et al., 2021). FTP is operationally estimated as 95% of a 20-min maximal effort. The approximate hierarchy is: CP ≥ FTP ≥ MLSS ≈ LT2 — with differences small in well-trained athletes but potentially meaningful in recreational cyclists.

Important caveat on LT2 precision (Valenzuela et al., 2021, Front. Physiol.). The D-max polynomial method shows good group-level agreement with Critical Power (mean bias ~0.5 W, Hedges' g = 0.01), but with wide individual limits of agreement (±52 W). This means that in any given athlete, measured LT2 may differ from true CP by 50 W or more — without either measurement being incorrect. Zone prescriptions should therefore be treated as a physiologically grounded starting point, to be refined using perceived exertion, heart rate response, and training performance data.

Zone model

Developed with Claude (Anthropic) as an AI coding assistant. Scientific content, methodology, and reference selection reflect the author's expertise. All 54 references were individually verified for DOI accuracy and bibliographic correctness.