The indoor-versus-outdoor argument keeps reappearing in cycling circles, usually in the form of one rider insisting Zwift is destroying outdoor riding and another rider insisting outdoor riders romanticise weather they would skip in any other context. Both are wrong. Indoor and outdoor training are not in competition. They develop different qualities, and the rider who uses both, deliberately, gets faster than the rider who picks a side and defends it.
This is the session-by-session split we use with Not Done Yet coaching members, framed around what actually wins in each environment. The question is never "indoor or outdoor in general." It is "for this specific session, on this specific week, given this specific rider's life — which one wins?"
What indoor training is genuinely good at
Indoor training, on a smart trainer integrated with a structured workout platform, has three durable advantages.
Precision. When ERG mode is holding the wattage to within five watts of target across a 4-minute VO2 effort, the rider doesn't have to manage the power. They manage the breathing, the cadence, the position, and the mental discipline. The signal-to-noise ratio of the workout is high. Outdoors, the same VO2 effort is constantly perturbed by terrain, traffic, wind, and the rider's own pacing — the file looks rougher even when the average is similar.
Consistency. Indoor training does not care about rain, ice, dark, traffic, or summer afternoons that hit 35 degrees. The riders who sustain training year-round across the climates of the UK, Ireland, and most of the US do so because the trainer is sitting there waiting for them on the days the road won't cooperate.
Time efficiency. A 60-minute indoor session is 60 minutes of training. A 60-minute outdoor session, accounting for getting kitted up, navigating traffic, finding clear road, and showering at the end, is 80-95 minutes of total time investment. For the rider with a 6:30am window before the kids wake up, the math is unforgiving.
These advantages compound. Most amateur cyclists who add the trainer to their toolkit, with a structured platform, gain meaningfully more from the consistency than from any other training change.
What indoor training is genuinely bad at
Three things, all real, all under-discussed.
Long rides. A 4-hour indoor ride is a different physiological experience from a 4-hour outdoor ride. The cadence variability, the terrain-driven pacing changes, the drafting and group dynamics on a real road, the gentle differences in fuelling timing as the environment shifts — none of it is fully present in even the best simulator. The aerobic stimulus is similar. The neuromuscular and pacing development is not.
Race-specific skills. Bike handling, descending, cornering at speed, working in a group, the timing of a paceline pull — all develop on the road or not at all. The rider who only trains indoors and shows up to a race fit but rusty is a familiar pattern.
Durability under fatigue. The pattern that shows up in coaching files is consistent. Indoor-only riders often have strong fresh power but poor power-at-fatigue at the third hour of a long ride. Outdoor base work develops a kind of durability that the trainer does not replicate, even at the same kilojoule cost.
These are not knockout arguments against indoor training. They are arguments against using indoor training for sessions where its weaknesses matter most.
What outdoor training is genuinely good at
The mirror image.
Aerobic durability. Long zone 2 rides outside, week after week, build the base that everything else stands on. Three to five hours of patient endurance riding, with the cadence variation and pacing nuance the road imposes, develops a quality the trainer cannot fully reproduce.
Race specificity. Group rides, paceline efforts, descending practice, real climbs at race pace — all require the road. No simulator yet replaces the experience of holding 50kph in a paceline of seven other riders, or the technical demand of a fast descent in a cross-wind.
Long-effort threshold and sweet spot work. A 40-minute climb at sweet spot, ridden outdoors with the natural pacing variation a real ascent imposes, develops something different from a 40-minute ERG-mode sweet spot interval indoors. The outdoor effort tends to ride better; the engagement of the environment makes the duration sustainable.
Recovery rides. A 90-minute easy spin outdoors, on a quiet road, with low traffic, on a good day — there is no equivalent indoors, and the mental recovery component is real.
What outdoor training is genuinely bad at
Short, high-precision intervals. VO2 work is the worst-case example. A 4-minute effort at 115% of FTP, ridden outdoors, gets perturbed by every junction, hill, and gust the road throws at it. The average power may be on target; the actual physiological stress is more variable than the file suggests. Indoors, the stimulus is clean.
Time-constrained sessions. A 45-minute session indoors is 45 minutes of training. A 45-minute session outdoors, after kit-up and travel time, is rarely possible. The rider with a tight midweek window is better served by the trainer.
Winter consistency. This is location-dependent. In the UK, Ireland, and most of the northern US, four months of the year impose dark, cold, wet riding that most amateurs will not consistently complete. The trainer is the reason their February doesn't disappear from the form curve.
The session-by-session split
This is the working default we run with the Not Done Yet coaching community. Local conditions and individual preferences shift the answer at the margin, but the structural pattern holds.
Long ride (3-5+ hours). Outside, almost without exception. The aerobic base, durability, fuelling practice, and pacing development of the long ride all benefit from the road environment. If weather genuinely makes outdoor riding unsafe — ice, severe storm, sub-zero with wind — substitute a 2.5-3 hour indoor ride at zone 2 with periodic over-threshold pushes to maintain neuromuscular engagement. This is a substitute, not an equivalent.
VO2 intervals (3-5 minutes at 110-120% of FTP). Inside, almost without exception. Indoor precision lets you actually ride at the prescribed wattage rather than averaging there. The shorter the interval and the higher the intensity, the more the indoor environment helps.
Short threshold (2 × 12-20 minutes at 95-100% FTP). Either, with a slight indoor preference. The clean execution of indoor ERG mode tends to produce a higher-quality file. Outdoor execution is fine on a stretch of road or a long enough climb that you don't have to stop.
Long threshold (3 × 20+ minutes, or 1 × 40 minutes at 95% FTP). Outside, on a sustained climb if possible. The natural pacing variability of the road keeps the effort sustainable in a way ERG mode at the same duration often doesn't. Most riders mentally tap out of long-duration ERG mode threshold sessions before the physiological stress catches up.
Sweet spot (3 × 15-25 minutes at 88-93% FTP). Either. The duration and intensity sit in a window where both environments work well. Some riders strongly prefer outdoor for sweet spot; others prefer indoor. Personal preference is a legitimate tiebreaker here.
Sub-threshold endurance (2-3 hours, mostly zone 2 with occasional surges). Outside, almost without exception. This is where the durability and pacing work of the trained amateur cyclist accumulates.
Recovery ride (45-60 minutes, zone 1). Outside if conditions allow; inside if they don't. Outdoor wins on mental recovery; indoor is acceptable if the alternative is skipping.
Time-crunched 45-minute hard session. Inside. Almost no outdoor option produces a 45-minute high-quality file when you account for the time cost of getting outside.
Race simulation (final 4 weeks of an event-specific block). Outside, on terrain that resembles the event. The fitness benefit is real and the psychological benefit is real.
Platform notes (in passing)
The platform you use indoors matters less than the trainer being there at all, but a few notes worth making.
Zwift wins on engagement for most riders, particularly through long winter blocks. The combination of group rides, race series, and visual variety makes the difference between four sessions a week and two.
TrainerRoad has the cleanest workout library and a useful adaptive-progression model. The integration with TrainingPeaks works well, and the AI FTP detection has improved markedly over recent versions.
SYSTM (the former Sufferfest) carries strong multi-month structured plans and named workouts that build a lot of cyclists' sense of consistency.
For the rider who runs structured TrainingPeaks plans — which is most riders in the Not Done Yet coaching community, given the partnership and the way Vekta plans push directly into TrainingPeaks — Zwift handles the workout file natively and the rider can ride the prescribed session without manually programming the trainer. This reliability is the practical reason most coached amateurs gravitate toward Zwift as the workout platform of choice.
How to think about the seasonal split
The total annual training of a UK or northeastern US masters cyclist might split roughly 60% outdoor and 40% indoor across the year. December through February is heavily indoor (75-90% indoor on most weeks). May through September is heavily outdoor (10-25% indoor, mostly for time-constrained midweek sessions or specific VO2 work). March-April and October-November are the transition months where the split becomes 50:50.
The ratio doesn't matter much. What matters is consistency — the rider who completes 90% of their planned sessions, regardless of how the indoor/outdoor split lands, beats the rider who completes 70% because they refuse to ride indoors and the weather is uncooperative for half the year.
Indoor and outdoor are not aesthetic choices. They are tools for different jobs. The rider who uses both gets faster than the rider who picks a side. The argument was always a false one.