The long session
When it is needed and when the week can be spread instead: for events up to about two hours it is optional, for 2-5 h events a few rehearsals before the race, beyond 5 h a regular part of the specific phase with the hard work late in it; no trial compares one long session with the same time split for long-event performance, so the duration bands are practice
Decision
| Event duration | The long session | Source and how sure |
|---|---|---|
| Up to ~2 h (TT, hill climb, crit, short road race, HYROX, 10 km) | Optional. Spread the hours however the week works; 90-120 min covers fuelling practice. Half marathon: a longest run over 21 km went with a faster finish and less fade (REF-1101 B, observational) | CP held to 80 min and fell 9% by 2 h (REF-145 B+); VT1 −5% at ~139 min (REF-143 B); same ~10% time-trial gain either arrangement (REF-1096 B, REF-1097 B+); band is practice |
| ~2-5 h (gran fondo, marathon, 70.3, long road race) | Spreading is fine in base. In the last 8-10 weeks 1-2 rehearsals of roughly 60-80% of race time with race fuelling and pacing. Marathon: a longest run of at least ~25 km | Practice; marathon: a longest run <25 km went with a slower finish (β 13.44, likely minutes; REF-1101 B, observational) |
| Over ~5 h (200-mile gravel, ultra, Ironman, marathon MTB) | A regular part of the specific phase: about 2-3 sessions of 50-80% of race duration in the last 8-10 weeks, one a full rehearsal, with hard work placed late (SYNTH-32, SYNTH-50) | Practice; the pro fade after ~4 h is real (REF-196 B); prolonged sessions with race-pace bouts are a suggested lever (REF-141 B+, perspective) |
Rules
Stacking costs quality. Intervals done 1 h after a 90-min steady ride ran at 297 W against 323 W done on their own day (REF-1097, B+). When the volume is spread or doubled, keep the key session first or on its own day.
One long session is a bigger and costlier stimulus than the same work split. Heart rate, lactate and RPE drifted up in one 60-min session, while heart rate and lactate fell in the second short session when it was split 6.5 h apart; the long version left more next-morning fatigue (REF-159, B, acute). Choose by purpose: drift and fatigued work → keep it whole; volume at low cost before a key day → split.
Durability has levers that need no long session: heavy strength twice a week (5-min power after a 185-min ride 371 → 400 W, REF-146 B+), carbohydrate during the ride (60 g/h cancelled the 2-h fall in CP, REF-145 B+), and in beginners even 1.6 h/week of hard riding improved it as much as 6.8 h of easy riding (REF-1099, B, untrained).
In elite riders cutting ~5 h/week of moderate-intensity riding kept fresh ~20-min performance, and the time trial after a 120-min preload did not improve significantly overall (p = 0.19; per-group results not reported) (REF-1100, B).
What only a long session gives (practice): pacing and fuelling rehearsed in the fatigued state, contact points and position at hour three and beyond, and a direct read of the athlete's own fade. A durability number is comparable only on a matched preload, fuelling and conditions (REF-1098, B).
If there is a long session, fill it rather than ride it flat (practice, SYNTH-32 §4): for a long event, easy hours first and race-specific work late, with targets below fresh zones. That is still a long session — it is how to use one, not a way to avoid one.
Signs the long session is needed now (practice): the event is longer than about 5 h or the athlete's first at that length; they fade late in races; decoupling is high on long rides (SYNTH-32 §3).
Evidence
Practical Application
Limitations
How this entry was checked and changed
(no training trial isolates the long session; the onset of decline and the distribution studies are B/B+; the marathon long-run association is observational; every duration band is practice)