REF-1304 · Devantier-Thomas et al., 2024 · European Journal of Applied Physiology · Paper · PubMed · Free full text

Will a leg session ruin tomorrow's ride?

In 10 trained cyclists new to lifting, a 10-km time trial showed no detectable change 24-48 h after heavy legs (test MDC 52 s); effort, breathing and early pace were worse (B).

Main caveat: No control group and n = 10 men; a null result for the time trial at this size is weak evidence of no effect.
limited evidenceMeasured on trained athletes

Devantier-Thomas and colleagues asked whether a second resistance-training bout protects cycling efficiency and performance better than the first.

Ten trained male cyclists with no lower-body resistance training in the previous six months did the same session twice, at least two weeks apart: Smith-machine back squat, single-leg press, leg extension and leg curl, 3 sets of 6 at 95% of their 6-repetition maximum (set to avoid failure), about 1 s up and 1 s down, 2 min rest.

The day before (baseline), 24 h and 48 h after each bout they measured creatine kinase, soreness and three jumps, then rode 10 min at 70% and 10 min at 90% of the second ventilatory threshold and, 5 min later, a 10-km simulated time trial. Soreness rose at 24 and 48 h after both bouts and was lower after the second (interaction p = 0.027).

Squat-jump height was lower at 24 and 48 h, countermovement jump at 24 h. During submaximal riding, ventilation at 70% of VT2 and perceived exertion at 90% of VT2 were higher at 24 and 48 h; oxygen uptake, heart rate, lactate and gross efficiency did not change.

Time-trial duration (about 934-949 s), average power, heart rate and perceived exertion did not differ by time or bout; power over the first 3 km fell from 281.1 W at baseline to 268.5 W at 24 h and 272.2 W at 48 h. Net efficiency at 90% of VT2 was higher in bout 2 than bout 1 (24.3 vs 23.8%), and cadence in the first 3 km was higher in bout 2.

The authors conclude that cyclists kept their time-trial times despite muscle damage, possibly by changing pacing and cadence, yet advise avoiding maximal or time-trial sessions for 48 h after at least the first two bouts, and sub-maximal cycling until after the second bout "to enhance cycling efficiency".

Our reading: support for a shorter lift-to-quality gap in cycling than in running — but the "no change" comes from a test whose minimal detectable change was 52 s in a ~15.6-min trial, and the higher effort and the slower start show the session still cost something.

Protocol

Participants: n = 10 healthy men; age 38 ± 13 y, height 180.4 ± 7.0 cm, body mass 80.1 ± 10.1 kg, VO2max 51.0 ± 7.6 mL·kg-1·min-1; recreational or competitive cyclists, 150-300 km a week.
Familiarity: resistance-untrained (no lower-body resistance training in the previous six months); familiarisation with the jumps and the cycling test in week 2.
Design (5 weeks): week 1, VO2max test and 6RM tests; week 2, familiarisation; week 3, bout 1 with tests the day before (Tbase), 24 h (T24) and 48 h (T48) after; at least two weeks later, bout 2 with the same tests. No control group.
Lifting bout: Smith-machine back squat, single-leg press, leg extension, leg curl; 3 × 6 at 95% of 6RM "to allow participants to complete each set without failure"; about 1 s concentric and 1 s eccentric; 2 min passive rest; 72 repetitions in total.
Cycling test: own bicycle on a Wahoo KICKR; 10 min at 70% and 10 min at 90% of VT2 (2 min rest between), VO2, VE, HR averaged over the last 5 min, RPE in the last 2 min, lactate after each stage; gross and net efficiency; 5 min later a 10-km flat simulated time trial (Zwift), with duration, power and cadence by segment (1-3, 4-6, 7-10 km) and RPE at 3, 6 and 9 km.
Markers: capillary CK (Reflotron; intra-assay CV 7.2%), soreness on a 1-10 scale during a full squat, countermovement, squat and drop jumps.
Controls: same shoes and bike, no caffeine within 2 h, no strenuous exercise for 24 h before tests, no recovery modalities, all sessions 5:00-7:00 am.
Reliability (two baselines): time trial ICC 0.99, SEM 18.9 s, minimal detectable change 52.35 s; countermovement jump ICC 0.94.
Analysis: two-way (bout × time) repeated-measures ANOVA, Bonferroni post hoc, alpha 0.05; a priori sample size of 10.
Funding and interests: funding: James Cook University (open access funding enabled and organised by CAUL and its member institutions); "The authors declare that they have no conflict of interest." Equipment makers (Wahoo, Zwift) are named as equipment only.

Key Results

Performance — 10-km time trial (Table 5; no interaction, time or bout effect):

Baseline24 h48 h
Duration, bout 1 (s)934 ± 33941 ± 39934 ± 39
Duration, bout 2 (s)944 ± 34949 ± 51942 ± 36
Average power, bout 1 (W)259 ± 35250 ± 35254 ± 37
Average power, bout 2 (W)254 ± 36251 ± 42254 ± 38
Power, 1-3 km (main effect of time)281.1 ± 34 W268.5 ± 30.13 W (p < 0.05)272.2 ± 31.28 W (p < 0.05)
Duration p = 0.884 (interaction), 0.50 (time), 0.21 (bout). Cadence in the first 3 km was higher in bout 2 (86.7 ± 6.0 vs 89.4 ± 6.85 rpm, p = 0.01).

Submaximal riding (Tables 3-4; main effects of time):

MeasureBaseline24 h48 h
Ventilation at 70% VT2 (L/min)72.01 ± 10.2074.48 ± 10.4573.76 ± 9.4
RPE at 90% VT215.25 ± 1.5416.40 ± 1.9516.45 ± 1.45
VO2, HR, lactate, gross efficiencyno main effect of time
Net efficiency at 90% VT2 (main effect of bout)bout 1 23.8 ± 1.2% vs bout 2 24.3 ± 1.0% (p < 0.05)

Markers (Table 2):

MeasureBoutBaseline24 h48 h
Soreness (1-10)11.4 ± 0.76.4 ± 2.06.3 ± 2.4
Soreness (1-10)21.3 ± 0.54.8 ± 1.24.5 ± 1.8
CK (U·L-1), no time or interaction effect1119.7 ± 46.7417.0 ± 293.3396.3 ± 606.2
CK (U·L-1)2133.3 ± 53.5240.3 ± 128.9150.2 ± 118.5
Squat jump, main effect of time (cm)both28.0 ± 5.725.5 ± 6.625.5 ± 6.5
Countermovement jump, main effect of time (cm)both28.3 ± 4.725.5 ± 5.5 (p = 0.01)25.1 ± 5.7 (p = 0.12)

Mechanisms

Measured: time-trial time held while power in the first 3 km fell and perceived exertion and ventilation during submaximal riding rose; in bout 2 the riders used a higher cadence early; jump height fell with no change in drop jump.
Authors' interpretation: cyclists with a cycling background compensate for muscle damage by changing cadence and pedalling strategy, which non-cyclists in earlier studies could not; the higher ventilation may come from group III and IV afferent signals linked to soreness; the efficiency gain in bout 2 reflects the repeated bout effect.
Inference: a slower start and a higher effort at the same output are the cost; a cyclist can still produce the planned average on a threshold day after a lift, but pacing-sensitive and maximal work (sprints, tests) is where the damage shows (REF-701).

Practical Application

Will a leg session ruin tomorrow's ride? In 10 trained cyclists new to lifting, a 10-km time trial showed no detectable change 24-48 h after heavy legs (test MDC 52 s); effort, breathing and early pace were worse (B).
For a cyclist, this study does not rule out a self-paced threshold ride 24 h after a lift — no detectable time-trial change in novices, though the first 3 km were slower.Not from this paper; the base's own practice: a familiar heavy lift ≥24 h before threshold or VO2max work and ≥48 h before a sprint or a test, a new or damaging one ≥48 h before quality — for cyclists "the base keeps 48 h after a new lift" (SYNTH-31).Sprints, power tests and ramp tests keep the longer gap, because peak power is what muscle damage cuts (REF-701; SYNTH-23).
The authors' own advice is stricter than their data: they recommend avoiding maximal or time-trial sessions for 48 h after the first two bouts despite no detectable change in time-trial time; their reason is session quality and pacing, which fits the measured fall in power over the first 3 km (281.1 to 268.5 W at 24 h, about 4.5%, our arithmetic).
Start a cyclist's strength block knowing the second session hurts less — soreness was lower at 24 and 48 h after bout 2 (this study; REF-1297); the higher net efficiency in bout 2 was already present before that lift, so it is not evidence of protection.

Limitations

No control group and n = 10 men; a null result for the time trial at this size is weak evidence of no effect.
Reporting that does not add up: the time trial's SEM of 18.9 s is inconsistent with its ICC of 0.99 given a between-rider SD of about 33-39 s (an ICC of 0.99 implies an SEM of about 3-4 s, our arithmetic); either way the reported minimal detectable change (52.35 s) is far larger than any observed change.The abstract reverses the net-efficiency values ("improved at Bout 2 (23.8 ± 1.2) than Bout 1 (24.3 ± 1.0%)"); the Results give bout 1 = 23.8%, bout 2 = 24.3%. The text puts the countermovement-jump time effect at p = 0.02 and the table at p = 0.03; the time-trial paragraph cites Table 4 for data in Table 5.
The time trial followed 20 min of submaximal riding, on a virtual flat course; outdoor, longer or hilly efforts not tested.
Resistance-untrained cyclists; cyclists who lift year-round may differ (likely less cost, inference).
Funding: James Cook University; no conflicts declared.
How this entry was checked and changed
Why this level

n = 10, no non-lifting control, read from the full text (Europe PMC, open access, CC BY 4.0). A real cycling performance outcome in real cyclists, but small, and the time trial's reported minimal detectable change (52.35 s) is large relative to the changes being looked for

Full text (checked 2026-10-27)
Ten healthy men (age 38 ± 13 years), recreational or competitive cyclists cycling between 150 and 300 km per week; any lower body resistance training within the preceding six months excluded (full text)
3 sets of 6 repetitions of Smith-machine back squat, single-leg press, leg extension and leg curls at 95% of 6RM to allow participants to complete each set without failure; a minimum of two weeks between bouts (full text)
Time-trial duration, bout 1: 934 ± 33, 941 ± 39, 934 ± 39 s; bout 2: 944 ± 34, 949 ± 51, 942 ± 36 s; interaction p = 0.884, time p = 0.50 (full text)
Average power 1-3 km decreased at T24 (268.5 ± 30.13 W) and T48 (272.2 ± 31.28 W) compared to baseline (281.1 ± 34 W) (full text)
VE at 70% VT2: T24 74.48 ± 10.45 and T48 73.76 ± 9.4 vs Tbase 72.01 ± 10.20 L.min−1; RPE at 90%: T24 16.40 ± 1.95 and T48 16.45 ± 1.45 vs 15.25 ± 1.54 (full text)
Net efficiency increased from Bout 1 to Bout 2 during the 90% stage (23.8 ± 1.2 vs 24.3 ± 1.0%) (full text)
DOMS lower during Bout 2 at T24 (p = 0.006) and T48 (p = 0.010); DOMS bout 1: 6.4 ± 2.0 and 6.3 ± 2.4; bout 2: 4.8 ± 1.2 and 4.5 ± 1.8 (full text)
Time-trial test-retest reliability: ICC 0.99, SEM 18.9, MDC 52.35 (full text)
Funding: James Cook University; Open Access funding enabled and organized by CAUL; the authors declare that they have no conflict of interest (full text)