ScienceMarch 28, 2026

We tested 297 cyclists. Here's what FTP gets wrong.

by Steinar Agnarsson

If you ride with a power meter, you almost certainly know your FTP. Functional Threshold Power is the number that defines you as a cyclist -- your training zones, your race pacing, your progress over time. Platforms like Zwift, TrainerRoad, and Wahoo SYSTM all build their entire coaching logic around it.

But FTP carries a hidden assumption that almost nobody questions: that every rider can sustain their lactate threshold power for approximately 60 minutes.

We tested 297 cyclists. That assumption is wrong for most of them.

The 60-minute doctrine

FTP was popularized by Andrew Coggan and Hunter Allen in the mid-2000s as a practical proxy for lactate threshold power. The idea was elegant: instead of requiring a lab test with blood draws, you could estimate your threshold from a hard 20-minute effort (multiply by 0.95) or, ideally, from a full 60-minute time trial.

The underlying claim is physiological: at your lactate threshold, lactate production and clearance are roughly balanced. This is the highest intensity you can sustain in a metabolic steady state. And according to the FTP framework, that steady state lasts about an hour.

This is where the trouble starts. The 60-minute figure is a population average. It is not a physiological law.

What we actually found

At Driftline, we analyzed data from 297 cyclists in the GoldenCheetah OpenData project -- a total of 13,887 ride sessions with power and heart rate data. For each rider, we fitted TrueZone's physiological model to extract their individual parameters, including their Lactate Threshold Power (LTP) and their endurance parameter E.

LTP in TrueZone corresponds to the V2 power -- the second ventilatory threshold in TrueZone's threshold system. This is the physiological boundary where lactate accumulation begins to accelerate and a metabolic steady state becomes unsustainable. It is the same boundary that FTP claims to estimate.

The critical question: how long can each rider actually sustain their LTP?

The answer depends almost entirely on their endurance parameter E. And E varies enormously across the population.

  • Riders with E = 0.4 (low endurance) could sustain LTP for roughly 30 minutes
  • Riders with E = 0.7 (average endurance) could sustain LTP for roughly 60 minutes
  • Riders with E = 0.9 (high endurance) could sustain LTP for roughly 95 minutes

FTP's 60-minute assumption only holds when E is approximately 0.70 -- which happens to be the population average. For everyone else, the number is wrong.

The scale of the error

This is not a minor calibration issue. Among our 297 cyclists:

186 riders had E greater than 0.75. For these cyclists, FTP underestimates how long they can sustain their threshold. Their actual threshold duration is longer than 60 minutes -- in some cases significantly longer. These are typically well-trained endurance athletes with efficient fat oxidation and high fatigue resistance. When a platform tells them their "threshold" pace is based on a 60-minute effort, it is selling them short.

38 riders had E less than 0.65. For these cyclists, FTP overestimates their threshold duration. They cannot hold their lactate threshold for 60 minutes -- some can only manage 30-40 minutes before accumulating unsustainable lactate levels. These tend to be riders with strong anaerobic capacity but less aerobic development. When they pace a 60-minute effort based on FTP, they blow up.

The remaining 73 riders fell in the sweet spot where FTP's assumption roughly holds. That is less than 25% of the sample.

Every zone inherits the error

Here is the part that matters for daily training. FTP is not just used to estimate one number. It is the foundation for an entire zone system. In Coggan's classic model:

  • Zone 1 (Active Recovery): less than 55% of FTP
  • Zone 2 (Endurance): 56-75% of FTP
  • Zone 3 (Tempo): 76-90% of FTP
  • Zone 4 (Threshold): 91-105% of FTP
  • Zone 5 (VO2max): 106-120% of FTP

Every one of these zones is defined as a percentage of FTP. If your FTP is wrong -- not because you tested poorly, but because the 60-minute assumption does not match your physiology -- then every zone is wrong.

For a high-endurance rider (E = 0.85), Zone 2 based on FTP might actually be too easy. Their true aerobic threshold is higher relative to their FTP than the model assumes. They end up spending long rides at an intensity that provides less training stimulus than intended.

For a low-endurance rider (E = 0.55), the opposite happens. Their "Zone 2" rides are actually too hard -- creeping into a metabolic intensity that accumulates fatigue without building the aerobic base they are targeting.

But there is a second, less obvious problem. Even if FTP were perfectly accurate, the fixed percentages used to define zones would still be wrong for most riders. Endurance does not just affect how long you can hold threshold -- it affects where your other thresholds sit relative to it. A high-endurance rider has their aerobic thresholds compressed closer to their lactate threshold, while a low-endurance rider has them spread further apart. The ratio between Zone 2 and Zone 4 is not fixed -- it depends on the geometry of your individual threshold system.

FTP-based zones assume the same percentages for everyone. TrueZone derives zone boundaries from each rider's individual threshold alignment. The result is two layers of error in FTP-based training: wrong threshold and wrong zone proportions.

The zones feel wrong because they are wrong. Not by a trivial amount -- by enough to change the nature of the training stimulus.

What this means for platforms

Zwift, TrainerRoad, Wahoo SYSTM, and dozens of other platforms have built sophisticated training systems on top of FTP-based zones. Structured workouts, adaptive training plans, race pacing strategies -- all of it flows from FTP.

This is not a criticism of those platforms. They have done remarkable work making structured training accessible. But they are building on a foundation that treats every rider as physiologically identical in their threshold durability. The data says otherwise.

The fix is not to abandon the concept of threshold power. The fix is to stop pretending that one parameter captures enough about a rider's physiology to define all their training zones.

What you actually need is a model that accounts for how a rider's thresholds relate to each other -- where their aerobic threshold sits relative to their anaerobic threshold, how quickly they fatigue above threshold, how efficiently they use fat as fuel at sub-threshold intensities. These are the things that the endurance parameter E captures.

The deeper issue

FTP became dominant because it was practical. You did not need a lab. You needed a power meter and 20 minutes of suffering. That practicality was genuinely valuable, and it moved the sport forward.

But practicality is not the same as accuracy. The field settled on FTP not because the science supported a universal 60-minute threshold duration, but because the science had not yet provided a better alternative that was equally simple to measure.

That alternative now exists. With continuous heart rate and power data from modern sensors, it is possible to fit a physiological model that captures individual variation in endurance, threshold alignment, and fatigue resistance -- not from a single test, but from the rides you are already doing.

The 297 cyclists in our analysis did not perform any special tests. They just rode. Their data told us everything we needed to know about their individual physiology, including how long they could actually sustain their threshold.

FTP was a useful simplification for its era. But the era of one-number-fits-all training zones should be ending. Your physiology is more individual than that, and the data to prove it already exists on your bike computer.