How to Do an FTP Test

Functional Threshold Power is the highest power you could, in theory, sustain for about an hour, and in practice it is the number every structured platform uses to scale its workouts: Zwift, TrainerRoad, Wahoo SYSTM, Garmin, all of them. Set it too high and every threshold interval becomes a failed interval. Set it too low and a season of sweet-spot work is a season of tempo. The test itself takes under an hour. Getting a number you can trust takes a little more thought, and most of that thought is about the equipment and the pacing rather than the protocol.

A note on scope: we have not run these protocols under lab conditions and we do not claim to have. The protocols and formulas below are Zwift's, documented on its support site and by Zwift Insider, and the 95% convention traces to the training literature that Zwift's 20-minute test follows. Where we describe how a test feels, that is the consensus of the coaching sources cited at the end, not our own bench data.

What has to be true before you test

Your power must be measured, not estimated. This is the only equipment requirement that actually matters. A trainer that estimates power from its resistance model can produce a number that drifts week to week, which turns a retest into noise. Look for a direct-drive trainer with a claimed accuracy of ±2% or better, or a power meter. We explain the measured-versus-estimated distinction in how accurate are smart trainers. A wheel-on trainer with a ±5% estimate will run the test, but do not treat the result as comparable to the last one.

Same device every time. FTP is only useful as a relative number. If you test on the trainer and train on power pedals, or test on the pedals and race on a different trainer, you are comparing two instruments. Pick one power source and use it for the test and the training. If you ride indoors and out, that argues for a power meter as the source on both — smart trainer vs power meter covers the decision.

ERG mode for the ramp, free ride for the 20. A ramp test is designed to be run in ERG mode: the trainer sets each step and you pedal until you cannot turn it. The 20-minute test is the opposite — Zwift disables ERG for the test block and delivers steady flat-road resistance, so you choose the gear and the effort. Make sure your trainer does both.

Calibrate, cool, fuel. If your trainer needs a spin-down, do it after ten minutes of riding, before the test, not cold — how to calibrate a smart trainer walks through it. Put a serious fan in front of you, because heat is the thing that ends a 20-minute effort early indoors. Test rested, fed and at the time of day you usually train.

The short answer

The gear that meets those conditions

#	Product	Best for	Score	Price
1	Wahoo Kickr Core 2 Measured power, proper ERG mode and WiFi at mid-range money — everything an FTP test needs from a trainer, and nothing it does not.	The trainer to test on	9.0★★★★★9.0 out of 10	$548.99 · View on Amazon #ad We may earn a commission · Price as of September 18, 2026
2	Favero Assioma Duo The fix for the most common FTP complaint — that the indoor number and the outdoor number disagree — is to measure both with the same device.	One number, indoors and out	9.2★★★★★9.2 out of 10	$538.99 · View on Amazon #ad We may earn a commission · Price as of September 18, 2026
3	Polar H10 Not required for the test, but the strap that makes the next test interpretable — the reference chest strap, with two Bluetooth channels so nothing misses it.	The heart rate cross-check	9.2★★★★★9.2 out of 10	$104.79 · View on Amazon #ad We may earn a commission · Price as of September 18, 2026

The ramp test

The ramp is the test to take if you have never taken one, and the one many coaches now recommend for everyone, because it removes pacing from the equation. After a warm-up the trainer holds a series of one-minute steps that rise by 20 W each — Zwift's standard ramp starts at 100 W — and you simply keep pedaling until the step you are on is one you cannot finish. Zwift then sets FTP at 75% of the highest one-minute average power you produced. That 75% figure comes from the observed relationship between maximal aerobic power and threshold in the coaching literature, which Zwift Insider traces to work in the late 1990s.

The ramp portion typically lasts five to twenty minutes depending on where you start and how fit you are. Riders under 60 kg or with an FTP below about 175 W should use Zwift's Ramp Test Lite, which starts at 50 W and rises by 10 W per minute; the finer steps give a more accurate result at lower power. Two things decide the quality of a ramp result: stay in ERG mode and do not shift, and do not stop early because it feels like it is about to get hard. It always feels that way two steps before the actual end.

The ramp's weakness is that it rewards riders with a strong anaerobic finish. If you can produce a big one-minute effort relative to your sustainable power, the 75% rule will overestimate your FTP, and your threshold intervals will feel impossible. If that describes you, take the 20-minute test as well and use the lower of the two.

The 20-minute test

Zwift's standard FTP test runs about 45 minutes in total: a warm-up that includes two short hard blocks to open the legs, a 20-minute free-ride effort, and a cool-down. Your FTP is 95% of your average power over the 20 minutes. Zwift turns ERG off for the effort, so the trainer behaves like flat road and pacing is entirely on you.

Pacing is the entire skill. Start at what you believe is your current FTP, or a little under, for the first five minutes. Reassess. If it feels sustainable, add a few watts every five minutes; if it does not, hold. The last three minutes are where you empty what is left. The classic failure is going out 10% too hard, fading at minute eight, and finishing with a 20-minute average lower than you could have held evenly. This is why the advice for a first test is to do a ramp first, then use that number to pace the 20 the next time.

Without Zwift

The protocols do not belong to Zwift. TrainerRoad, Wahoo SYSTM and Garmin each offer their own ramp or timed test built on the same principles. On any smart trainer with ERG mode you can build a ramp as a custom workout — one-minute steps rising 20 W — and on any trainer at all you can ride 20 minutes hard and take 95% of the average from your head unit. What you cannot do without measured power is get a number worth writing down, which is why the equipment section came before the protocols.

What to do with the number

Enter it into every platform you use, and enter the same number everywhere. Zwift will also auto-detect a higher FTP from any effort between roughly eight and sixty minutes, which is convenient but means your zones can shift mid-block without a formal test; decide whether you want that. Retest every six to eight weeks, at the end of a training block, on the same equipment, at the same time of day, with the same protocol. A retest on a different trainer or a different protocol is a new baseline, not a comparison.

And if the number is lower than you hoped, that is fine. FTP that is honest makes every workout that follows the right difficulty; FTP that is flattering makes them all wrong. The trainer does not care. Our best smart trainers roundup covers the units that measure to ±1% if repeatability is what you are after.