Does the first kilometre off the bike decide the run?
In 10 national-level male triathletes, opening 5% under fresh 10-km pace beat +5% (150 s) and −10% (59 s) over the 10 km. Triathlon, not HYROX.
Main caveat: No self-paced triathlon. The authors treat +5% as the athletes' usual strategy from race data, and the −5% is compared with imposed openings only.
Ten national-level male triathletes ran a free-paced 10-km control run and then three solo Olympic-distance time trials in random order, 10 days apart, with swim and bike speeds imposed as identical to their first triathlon.
Over the first run kilometre they followed an audio signal at 5% faster (+5%), 5% slower (−5%) or 10% slower (−10%) than control-run speed, then ran the remaining 9 km as fast as possible. The −5% opening gave the fastest run, 1.4% off the control run and not significantly different from it; +5% was 8.9% off and −10% 4.4% off.
The fast opening raised VO2, ventilation and heart rate at km 1, and speed then fell sharply through km 3. The −10% opening left VO2 and ventilation lower than −5% at 5 and 9.5 km and lactate lower at the end (at 5 km the Results give p = 0.12), which the authors read as not reaching the workload the athletes could have sustained.
They had hypothesised that a fast opening would be best, and the data went the other way. Our reading: a clean test of three openings in fit men, with a large effect for its size; it does not show −5% beats what they would choose alone, since no triathlon was self-paced.
Protocol
Who: n = 10 male triathletes, national level, at least four years of training and racing. 178 ± 5 cm, 68.2 ± 6.7 kg; 12.5 ± 1.9 km swimming, 220 ± 42 km cycling and 65 ± 12 km running a week.
Order: maximal running test; 10-km control run (C-Run, always first of the performance tests, free pace, 340-m indoor track, feedback only on kilometres completed); then three triathlons in randomised order with a 10-day recovery between them. Training load controlled and similar between tests.
Triathlons: 1.5-km swim in a 50-m pool, 40 km on a 166-m indoor cycling track, 10-km run on the same 340-m track, alone, 18-20°C. Swim and bike speed in all three matched the first triathlon, itself done as fast as possible, with an audio pacer.
First kilometre: 5% faster, 5% slower or 10% slower than mean C-Run speed, with a signal every 25 m. Then as fast as possible with distance feedback only.
Measures: VO2 and ventilation by portable analyser at 0.5-1, 4.5-5 and 9-9.5 km (subjects were stopped after 35 km of cycling to fit it, as the paper describes), heart rate, lactate at the end of the ride, at 5 km and at 10 km. RPE was not measured.
Analysis: two-way repeated-measures ANOVA, Newman-Keuls post hoc, p < 0.05. No sample-size calculation reported.
Funding and interests: technical support from INSEP, the French Federation of Triathlon and the University of South Toulon-Var; no conflict of interest declared.
Key Results
| Measure | C-Run | Tri-Run +5% | Tri-Run −5% | Tri-Run −10% |
|---|---|---|---|---|
| First km (s) | 200 ± 15 | 190 ± 14 | 210 ± 17 | 220 ± 18 |
| 10-km run (s) | 2000 ± 72 | 2178 ± 121 | 2028 ± 78 | 2087 ± 88 |
| Mean speed (km/h) | 18.0 ± 0.6 | 16.5 ± 0.9 | 17.8 ± 0.4 | 17.2 ± 0.6 |
| Off the control run | 8.9% | 1.4% | 4.4% | |
| Swim + bike (s) | 1278 + 4260 | 1275 + 4255 | 1281 + 4263 | |
| Overall time (s) | 7716 ± 196 | 7558 ± 188 | 7631 ± 191 | |
| VO2 at km 1 / km 9-9.5 (mL/min/kg) | 61.5 / 64.4 | 66.1 / 55.1 | 60.9 / 68.1 | 55.1 / 60.6 |
| Heart rate at km 1 (beats/min) | 176.2 | 186.0 | 168.0 | 159.0 |
| Lactate at 5 km / 10 km (mmol/L) | 3.2 / 3.4 | 4.9 / 4.1 | 3.8 / 5.4 | 3.6 / 4.0 |
Significance: −5% was faster than +5% (p = 0.005) and −10% (p = 0.02), with no significant difference from C-Run (p = 0.58). Swim and bike did not differ between triathlons (p > 0.05). The 20-s difference on the first kilometre between +5% and −5% went with a difference of 150 ± 21 s over the 10 km; between −5% and −10%, 59 ± 11 s.
What the fast start did: the remaining 9 km ran 2.7 ± 0.4 km/h slower than the first kilometre; after −5% they ran 1.0 ± 0.3 km/h faster. Speed fell 1.5 km/h per km between km 2 and 4, then 0.2 km/h per km over the next five, and there was an end-spurt.
Our arithmetic: the −5% first km (210 s) was slower than that run's average (2028 s over 10 km, 202.8 s/km), so the best opening was also a negative split. "Overall time" equals the three legs added up, so transitions are not in it.
Ventilation: at the end of km 1, 93.9 ± 6.1% of maximal ventilation at +5%, against 86.5 ± 7.8% at −5% and 81.7 ± 6.9% at −10%.
Mechanisms
Measured: higher VO2, ventilation and heart rate at the end of km 1 after +5%, and 4.9 ± 0.5 mmol/L lactate at 5 km despite a slower speed (16.4 vs 18.0 km/h for −5%). VO2, ventilation, heart rate and final lactate were highest late in the −5% run.
Authors' interpretation: after a fast opening the pace is centrally down-regulated to protect against premature fatigue, with metabolic and ventilatory strain as signals; they speculate RPE exceeded what a "central controller" allowed, and read the end-spurt as a sign of reserve. RPE was not measured.
Authors' interpretation: the −10% opening was furthest from the athletes' habits, so its unusual internal feedback may have stopped them finding their pace; more practice might help. They ask for further studies.
Inference: the lower speed and physiology after −10% fit under-pacing, but how much is habit, and how much would be lost after practice, was not tested.
Practical Application
Does the first kilometre off the bike decide the run? In 10 national-level male triathletes, opening 5% under fresh 10-km pace beat +5% (150 s) and −10% (59 s) over the 10 km. Triathlon, not HYROX.
Good for: the shape of the error. Both cost: the +5% opening, 10 percentage points from the best one, lost 150 s; the −10% opening, 5 points from it, lost 59 s (about 15 and 12 s a point, our arithmetic), so the asymmetry is smaller than the raw gap. "Slower is safer" holds only up to a point.
Not good for: a percentage for HYROX. The anchor (fresh 10-km pace), the 40-km ride before it and the 10-km run after it do not exist there.
Authors' advice: automate the −5% pace in back-to-back cycle-run training. Tested only as a race-day pace, not as a thing to train.
HYROX (inference): not from this paper; the base's own practice: first 1 km at or slightly under what you can hold for all eight (SYNTH-71). This trial is an analogue for the shape only: its first kilometre followed a swim and a ride, while run 1 of HYROX follows nothing, so neither the direction nor the −10% floor is tested there.
Limitations
No self-paced triathlon. The authors treat +5% as the athletes' usual strategy from race data, and the −5% is compared with imposed openings only. So "reduce their freely-chosen pace", in their conclusion, was not tested in this design.
Three steps tested. The best of −5%, −10% and +5% is not the optimum; 0% or −2% was not tried.
n = 10 men, national level, not "elite" as the conclusion says, on indoor tracks, alone, with no wetsuit, heat, wind or rivals. The fixed-order C-Run is not a randomised arm.
Interruption and equipment. Subjects were stopped after 35 km of cycling to fit a portable gas analyser, and ran wearing it.
No blinding reported, and the athletes must have known each opening from the pacer; expectation can move a self-paced run (our inference). The large −10% cost may partly reflect unfamiliarity (authors' own suggestion).
Reporting. The Discussion calls all three runs significantly slower than C-Run (p < 0.05), while the Results give p = 0.58 for −5%. The Discussion gives a 35-s gap between the +5% and −10% first kilometres, while the Results give 190 and 220 s, 30 s (our arithmetic). The abstract and Results agree on the headline times.
How this entry was checked and changed
Why this level
crossover with n = 10 (rubric B: <15), a performance outcome; read from the authors' accepted manuscript at HAL (hal-01724323)
Full text (checked 2026-10-30)
Time to complete the first 1 km: 200 ± 15 s, 190 ± 14 s, 210 ± 17 s and 220 ± 18 s for C-Run, Tri-Run+5%, Tri-Run-5% and Tri-Run-10% (full text)
10-km run: 2028 ± 78 s, 2000 ± 72 s, 2178 ± 121 s, 2087 ± 88 s for Tri-Run-5%, C-Run, Tri-Run+5%, Tri-Run-10%; p = 0.005 and p = 0.02 vs +5% and -10%, p = 0.58 vs C-Run (full text)
Overall times: 7716±196 (Tri-Run+5%), 7558±188 (Tri-Run-5%), 7631±191 (Tri-Run-10%) s (full text)
A 20 sec-variation in running time over the first kilometre led to a significant difference of 150 ± 21s; the differential time between Tri-Run-10% and Tri-Run-5% reached 59 ± 11s (full text)
Triathletes increased their 10-km C-Run time by 1.4%, 4.4% and 8.9% during Tri-Run-5%, Tri-Run-10% and Tri-Run+5% (full text)
Training 15 ± 3 h.wk-1, running VO2max 69.1 ± 7.1, ten male triathletes at a national level (full text)