REF-1325 · Inglis et al., 2020 · International Journal of Sports Physiology and Performance · Paper · PubMed
Is my FTP my steady-state threshold?
Not quite: in 18 trained cyclists MLSS was 93% of FTP (95% of 20-min power), from 8 W above to 51 W below it, and in 10 retested FTP missed a 12-W MLSS gain over a season (B).
Main caveat: Small, one lab. Eighteen riders (6 women) for the comparison and 10 (1 woman) for the season part, which had no control group.
Inglis, Iannetta, Passfield and Murias asked whether 95% of a 20-min maximal effort (FTP95%) gives the power at MLSS, and whether the 20-min test tracks a change in fitness.
Eighteen trained to well-trained cyclists (12 men, 37 ± 6 y; 6 women, 28 ± 6 y) did a ramp test, 2-3 constant-load 30-min trials to find MLSS (lactate rising ≤ 1.0 mmol/L between minutes 10 and 30) and a 20-min maximal effort, all in the lab.
MLSS power was 88.5 ± 4.8% of the 20-min power and 93.1 ± 5.1% of FTP95%; both FTP values were higher than MLSS (p < 0.05) and strongly correlated with it, but the bias was −17 W with limits of agreement of 9 to −44 W and individual differences from −8 to +51 W.
At MLSS heart rate was 162 ± 8 beats/min, RPE 5.0 of 10 and end lactate 4.3 mmol/L; at the end of the 20-min effort 175 ± 8 beats/min, 8.4 and 12.3 mmol/L.
Ten riders repeated everything at the end of a 7-month season: MLSS rose 12 ± 8 W (range +2 to +28 W) and VO2 at MLSS rose, while neither the 20-min power nor FTP95% showed a mean increase (p = 0.75) and VO2max showed no significant change (p = 0.45) — no evidence of a change in 10 riders, not evidence of none.
Our reading: FTP95% overshot the steady-state ceiling by about 7% on average (our arithmetic from 93.1%) and by up to about 50 W in one rider, and a season's physiological gain did not show up in it; zones anchored on FTP therefore sit a little higher against physiology than their labels suggest, and a flat FTP does not prove flat fitness.
Ten riders without a control group is thin evidence for the second point.
Protocol
Participants: n = 18 (12 men: 37 ± 6 y, 180 ± 6 cm, 79 ± 8 kg; 6 women: 28 ± 6 y, 171 ± 6 cm, 68 ± 9 kg), "trained to well-trained"; University of Calgary ethics board. Ten returned for the season retest (9 men, 1 woman; 39 ± 5 y); sample size n = 10 from the differences in a similar study.
Setting: electromagnetically braked ergometer (Velotron Dynafit Pro), about 21 °C; 4-5 sessions 48-72 h apart at the same time of day; no vigorous exercise the day before; similar diet across testing.
Ramp test: 4 min at 50 W, then 30 W/min (men) or 25 W/min (women) to exhaustion; VO2max = highest 30-s rolling mean; also used to predict the first MLSS load.
MLSS: 30-min constant-power trials after 4 min at 80 W; MLSS = the highest power with lactate stable (Δ ≤ 1.0 mmol/L) between the 10th and 30th minute; trials repeated until the criterion was met; riders blinded to power and time.
20-min test (FTP20): after 8 min at 80 W, the highest average power over 20 min; riders controlled the gearing, saw time and cadence but not power, no verbal encouragement; FTP95% = 95% of the 20-min mean.
Measures: breath-by-breath gas exchange (Quark CPET), heart rate (Garmin), finger-prick lactate every 5 min (Lactate Scout), RPE 0-10, feeling scale and felt-arousal scale before and after sessions; pacing as power in 5-min segments.
Season part: PRE testing two months before the racing season (mostly long endurance rides), POST at the end of a seven-month season; the riders trained 5-6 days a week, about 1.5-4 h per session.
Statistics: paired t-tests, repeated-measures ANOVA, Bland-Altman, Pearson correlation; p < 0.05.
Funding and interests: "This study was supported by funding from NSERC (RGPIN-2016-03698)"; an NSERC Master's Award to E.C. Inglis; Heart and Stroke Foundation of Canada National New Investigator Award to J.M. Murias; "The authors report no conflict of interest." No device maker, federation or data platform named.
Key Results
| Measure (full group, n = 18) | At MLSS | At the end of the 20-min test |
|---|---|---|
| Power relative to the 20-min test | 88.5 ± 4.8% of FTP20; 93.1 ± 5.1% of FTP95% | — |
| Heart rate (beats/min) | 162 ± 8 | 175 ± 8 |
| RPE (0-10) | 5.0 ± 1.7 | 8.4 ± 1.4 |
| End lactate (mmol/L) | 4.3 ± 1.2 (rise 0.7 ± 0.3 from min 10 to 30) | 12.3 ± 2.6 |
Agreement: FTP20 and FTP95% both higher than MLSS (p < 0.05); strong correlation between MLSS and FTP95%; bias −17 W, limits of agreement 9 to −44 W, individual differences −8 to +51 W (FTP95% minus MLSS). Absolute watts are in Table 1, not in the repository copy.
Ramp test: VO2max 4.00 ± 0.68 L/min; peak power 394 ± 67 W.
Season (n = 10): MLSS +12 ± 8 W (range +2 to +28 W; p = 0.002 in the abstract, "p=0.00" in the Results); VO2 at MLSS 3.63 ± 0.51 → 3.77 ± 0.51 L/min (+0.14 ± 0.13 L/min, p = 0.01); 20-min power change −18 to +26 W and FTP95% −17 to +25 W across riders, no mean increase (p = 0.75); VO2max 4.32 ± 0.53 → 4.37 ± 0.60 L/min (56.6 ± 4.3 → 57.7 ± 7.9 mL/kg/min, p = 0.45); MLSS lactate rise 0.7 ± 0.3 mmol/L at both times; pacing unchanged (p = 0.48-0.96).
Ratio at PRE and POST in the 10: 88 ± 6% (bias −20 W) at PRE and 92 ± 5% (bias −9 W) at POST — printed as "FTP95% represented … of PO at MLSS"; see Limitations.
Our arithmetic: MLSS at 93.1% of FTP95% means FTP95% is about 107% of MLSS on average.
Mechanisms
Measured: the 20-min effort ended at a lactate of 12.3 mmol/L and a heart rate 13 beats/min higher than MLSS, so the test is well above the steady state; MLSS power and its VO2 rose over the season without any rise in 20-min power.
Authors' interpretation: a 20-min maximal effort draws on both oxidative and non-oxidative energy, in proportions that differ between riders, so a fixed 95% (or 88%) cannot land on MLSS for everyone; as a performance test it also depends on motivation, which was not measured (feeling and arousal scales and pacing did not change). Exercising only 10 W above MLSS sharply shortens time to exhaustion (their earlier work), so the mismatch matters for training.
Inference: a rider with a large anaerobic contribution gets an FTP further above MLSS, so every %-of-FTP zone sits higher against that rider's physiology; and because the FTP–MLSS gap changed within riders over a season (bias −20 → −9 W), %-of-FTP zones drift against physiology even when FTP is "fresh".
Practical Application
Is my FTP my steady-state threshold? Not quite: in 18 trained cyclists MLSS was 93% of FTP (95% of 20-min power), from 8 W above to 51 W below it, and in 10 retested FTP missed a 12-W MLSS gain over a season (B).
Set "threshold" sessions a little under FTP when a steady state is meant: MLSS averaged about 7% below FTP95% here (our arithmetic), and the authors found the individual spread too wide to use any single percentage, 88% of the 20-min power included. This agrees with SYNTH-53 (FTP20 is not "an hour") and REF-747 (median time to exhaustion at FTP20 35-51 min).
From % of FTP to the first threshold, as arithmetic across studies: combining this paper's 93.1% with VT1 at about 78% of MLSS in other well-trained riders (REF-1323) puts VT1 near 73% of FTP for an average trained rider; a ride at 65% of FTP is then about 89% of VT1 and one at 50% about 68% (our arithmetic on group means of different samples). Individual riders can sit well away from this.
A flat FTP is not proof of flat fitness: over a season MLSS rose 12 W on average while the 20-min test showed no mean increase (p = 0.75; changes −18 to +26 W); read training data (heart rate at a fixed power, decoupling, interval progressions) before concluding a block failed (SYNTH-53). Ten riders and no control group, so this is a warning, not a rule.
Not from this paper; the base's own practice: easy riding is capped at about 70% of FTP (SYNTH-70); by the arithmetic above that sits a little below an average trained rider's VT1.
Limitations
Small, one lab. Eighteen riders (6 women) for the comparison and 10 (1 woman) for the season part, which had no control group; the 20-min test was done on a lab ergometer, blinded to power, which may differ from a road or trainer FTP test.
Version read: the authors' accepted manuscript; the published version may differ in wording, and Tables 1-3 (absolute watts for MLSS, FTP20 and FTP95%) are not in the repository copy, so the absolute powers could not be checked.
Reporting (things that do not add up): the season ratios are printed as "FTP95% represented 88 ± 6% of PO at MLSS" (PRE) and "92 ± 5%" (POST), which cannot be — FTP95% was above MLSS; with the stated biases (−20 and −9 W) they would fit MLSS as a share of the 20-min power (as in the full group's 88.5%) if MLSS was roughly 240-290 W, but this copy prints no MLSS watts, so which quantity was meant cannot be confirmed (our arithmetic).The sign convention also flips between "differences ranging from -8 to +51W" and "LOA = 9 to -44W". MLSS's season change is "p=0.00" in the Results and p = 0.002 in the abstract.
What FTP means here: only the 20-min test and its 95% were tested; 60-min, ramp-derived and software-estimated FTP values are not covered.
Funding and interests: public Canadian research funding (NSERC; Heart and Stroke Foundation of Canada); no conflicts declared.
How this entry was checked and changed
Why this level
a validation study in one laboratory with n = 18 and a 10-rider uncontrolled longitudinal part (the rubric's B for validation studies; single-lab, small n). Read from the authors' accepted manuscript in the Kent Academic Repository (found via OpenAlex); its Tables 1-3, which hold the absolute watts, are not in that copy
Full text (checked 2026-10-28)
Eighteen participants (12 males- 37± 6 years; 6 females- 28 ± 6 years); 10 participants returned to repeat the test series after 7-months of training (full text)
PO at FTP20 and FTP95% was greater than that at MLSS, with the PO at MLSS representing 88.5 ± 4.8% and 93.1 ± 5.1% of FTP and FTP95%, respectively (full text)
The calculated FTP95% overestimated PO corresponding to MLSS (i.e., bias = -17W) with differences ranging from -8 to +51W; LOA = 9 to -44W (full text)
Mean HR and RPE was 162 ± 8 bpm and 5.0 ± 1.7 at MLSS and 175 ± 8 bpm and 8.4 ± 1.4 at FTP20; mean end [BLa] was 4.3 ± 1.2 mmol∙L-1 for MLSS and 12.3 ± 2.6 mmol∙L-1 for FTP20 (full text)
MLSS was greater at POST compared to PRE (+12 ± 8W; range +2 to 28W); no increase was seen in mean PO at FTP20 (range -18 to +26W) and FTP95% (-17 to +25W) (p=0.75) (full text)
VO2max PRE 4.32 ± 0.53 L∙min-1, 56.6 ± 4.3 ml∙kg∙min-1, POST 4.37 ± 0.60 L∙min-1, 57.7 ± 7.9 ml∙kg∙min-1 (p=0.45) (full text)
At PRE, FTP95% represented 88 ± 6% of PO at MLSS (bias = -20W) whereas at POST, FTP95% values represented 92 ± 5% of PO at MLSS (bias -9W) (full text)
These ten participants trained on average 5-6 days per week, for ~1.5-4 hours per session (full text)
This study was supported by funding from NSERC (RGPIN-2016-03698); Heart and Stroke Foundation of Canada National New Investigator Award (# 1047725); the authors report no conflict of interest (full text)