REF-1729 · Forster et al., 2026 · Journal of Sleep Research, · Paper · PubMed

Seven definite evening-type cyclists rode a 20-km laboratory time trial 39.9 s faster at 18:00 than at 06:00 (about 2%) and felt least mentally ready at 06:00, while ten definite morning types showed no clear time-of-day difference, in a pilot crossover of 17 trained male cyclists and triathletes; the effect sizes were small (Cohen's d 0.23 for that contrast). Read in full

Main caveat: 17 riders against 38 needed by the authors' power calculation; 7 evening types; the authors call it a pilot and ask for replication.
limited, mixed evidenceMeasured on trained athletesRead in full

Forster and colleagues asked whether a cyclist's chronotype changes when in the day they ride a 20-km time trial best. Of about 70 competitive male cyclists and triathletes who completed the Morningness-Eveningness Questionnaire (the abstract says 76), only those classed as definite morning types (10) or definite evening types (7) were tested.

Each rode four self-paced 20-km time trials on an ergometer at 06:00, 12:00, 18:00 and 22:00 in counterbalanced order, after a standard submaximal warm-up, rating mental readiness, sleepiness and motivation beforehand.

Across all 17 riders finishing time varied with time of day (p = 0.008; the abstract gives a 2.1% ± 3.8% evening advantage), with the fastest mean at 22:00 (31:35) and the slowest at 06:00 (32:12), while mean power, heart rate, oxygen uptake and perceived exertion did not differ by time of day.

Judged against each rider's own average (a z-score), evening types were better at 18:00 than at 06:00 (p = 0.02), about 40 s faster; morning types showed no significant time-of-day difference. Evening types rated mental readiness lowest at 06:00, reported more sleepiness at 06:00 than morning types and rated exertion higher than morning types at every time.

Pacing did not change with time of day or chronotype. The authors conclude that evening types perform better later in the day and that this matters for scheduling training and preparing for morning competition; they call the result a pilot that needs replication.

Participants

n = 17 men (10 definite morning types, 7 definite evening types), from about 70 screened (abstract: 76).
Age 37.9 ± 10.3 y (22-55), body mass 77.9 ± 7.6 kg, height 181.0 ± 4.9 cm, training 9550 ± 2924 km/y, peak incremental power 356 ± 45 W (4.6 ± 0.5 W/kg), habitual total sleep 8.7 ± 0.7 h.
Inclusion: > 2 years of cycling competition, injury-free, no sleep disorder, no shift work or travel across time zones in the past month, habitual bedtime 22:00-23:30 and rising 06:00-07:30, caffeine ≤ 150 mg/day, no stated preference for a training time. The inclusion criteria list an age of 18-30 years, which the reported age range does not match.

Protocol

Visit 1: incremental ergometer test (50 W every 3 min) for peak power.
Visits 2-5: 20-km flat self-paced time trials at 06:00, 12:00, 18:00 and 22:00, randomised and counterbalanced, 2-7 days apart; no familiarisation trial (the authors cite earlier work showing stable times across three 20-km trials).
Standard warm-up: Lamberts and Lambert submaximal cycling test (5 min at 60%, 6 min at 80% and 3 min at 90% of HRmax), then 5 min rest.
Riders agreed to sleep 22:30-06:30 during the study and lived normally between sessions; no strenuous exercise for 48 h before each trial; training and nutrition, caffeine included, copied from the first trial by diary. How early riders woke for the 06:00 trial is not reported; the authors acknowledge that it may have shortened their sleep.
Laboratory 19 °C, fan, visual feedback of distance, power, cadence and heart rate; standard encouragement in the last split.
Measures: finishing time, mean power per 2 km, heart rate, gas exchange, peak blood lactate, RPE (0-10) every 2 km; mental readiness, sleepiness and motivation on 0-100 scales before each trial.
Core temperature by pill was planned but usable data came from only five riders and was not analysed.

Key Results

Finishing time (min:s, mean ± SD):

Time of dayAll (n = 17)Morning types (n = 10)Evening types (n = 7)
06:0032:12 ± 2:2231:59 ± 2:0332:30 ± 2:54
12:0032:07 ± 2:2231:53 ± 1:5632:26 ± 3:01
18:0031:59 ± 2:0832:04 ± 1:5131:15 ± 2:38
22:0031:35 ± 2:3331:19 ± 2:3131:58 ± 2:46
Time of day, all riders: main effect on finishing time F(3, 59) = 4.3, p = 0.008; abstract: evening better than morning by 2.1% ± 3.8%. No time-of-day effect on mean power (250.5-258.1 W; p = 0.97), heart rate (p = 0.78), RPE (p = 0.66), VO2 (p = 0.79) or RER (p = 0.53).
Chronotype: no main effect on finishing time (p = 0.95) or mean power (p = 0.72); RPE higher in evening types (main effect p < 0.001; e.g., 6.9 vs 6.1 at 06:00, 7.3 vs 6.5 at 22:00).
Individual z-score (time × chronotype): F(1, 60) = 5.13, p = 0.02; evening types better at 18:00 than 06:00 (p = 0.02).
Mean difference 06:00 minus 18:00 (Table 3): evening types 39.9 ± 30.9 s (Cohen's d 0.23); morning types 40.5 ± 75.9 s (d 0.04, not significant). Other contrasts d −0.29 to 0.26.
Perceptions: interaction for mental readiness (p < 0.01), sleepiness (p = 0.01) and motivation (p = 0.04); evening types least ready at 06:00 (vs 18:00 and 22:00), sleepier at 06:00 than morning types (medians 8.0 vs 2.4, arbitrary units) and more ready than morning types at 22:00; no time effect on motivation in post hoc tests.
Best time of day: all evening types had their best ride in the second half of the day (the Discussion says 6 of 7); 6 of the morning types (reported as "6 out of 11", although there were 10) had theirs in the first half.
Pacing: no effect of time of day (p = 0.77) or chronotype (p = 0.74).

Practical Application

Evening-type cyclists rode a 20-km time trial about 40 s (2%) faster at 18:00 than at 06:00 and felt least ready at 06:00; morning types showed no clear time-of-day difference. One pilot of 17 men with extreme chronotypes; the effect sizes were small.
This supports the direction of SYNTH-07's row for a late chronotype facing a morning race or session (late types worse early, REF-606) and its default of placing hard sessions where the athlete's day and chronotype allow. It adds a cycling time trial in trained men to the base's two studies in healthy volunteers. It does not support the idea that morning types peak in the morning: their times did not differ significantly.
For all riders the slowest mean time was at 06:00 and the fastest at 22:00, with power, heart rate and RPE no different; the 06:00 rides also probably meant an earlier alarm, which the authors did not measure.
Inference, not tested here: for an evening type, an early race or key session is the case to prepare for (wake-time shift, extra warm-up time), as SYNTH-07 lays out; this paper tested no countermeasure.
Inference, not tested here: evening types rated the same work as harder at every hour, so their RPE runs higher than a morning type's for the same ride.

Limitations

17 riders against 38 needed by the authors' power calculation; 7 evening types; the authors call it a pilot and ask for replication.
Internal inconsistencies: the Results text says performance was better at 06:00 than 22:00, while the table and abstract show 22:00 faster; the abstract says 76 screened, the Methods 70; the inclusion age of 18-30 y does not match the 22-55 y sample; the text says all evening types and the Discussion 6 of 7 had their best ride late; "6 out of 11" morning types for a group of 10; the evening-type table means differ by 75 s between 06:00 and 18:00, while the reported mean paired difference is 39.9 s; the z-score is described as built from power, but its signs follow time; Table 3's note says a negative difference means faster later in the day, yet the evening types' 06:00-18:00 difference is positive (39.9 s) while the text calls it 40 s faster at 18:00; the abstract's summary line says morning types peak earlier, while the Results found no significant time-of-day difference for them; the Discussion says riders may have woken earlier for the 06:00 trial, while the Limitations say wake-up times were kept the same whatever the testing time.
Finishing time differed by time of day while mean power did not (p = 0.97), which the paper does not explain.
Small effect sizes by the authors' own Cohen's d (0.23 for the headline contrast).
Wake time before the 06:00 trial and sleep the night before each trial were not measured; the 06:00 effect may partly be an early-alarm effect.
Men only, extreme chronotypes only, no familiarisation trial; core temperature data unusable.
No funding received; the authors declare no conflicts of interest.
How this entry was checked and changed
Why this level

the evening-type claim rests on a within-person contrast in 7 riders and a between-group comparison of 7 against 10 (B by design: crossover under 15), lowered one step because the sample is 17 of the 38 the authors' power calculation required and the text, tables and abstract contradict each other on several points (see Limitations)