Zwift FTP Calculator and Training Zones

Functional threshold power is the highest power you can hold in a quasi-steady state for roughly an hour. Andrew Coggan, who introduced the term to cyclists, was blunt about how to find it: "probably the easiest and most direct way of estimating a rider's functional threshold power is to simply measure their average power during a ~40 km (50-70 min) TT." Every other protocol on this page is a shortcut to that hour, and every multiplier is an admission that the shortcut is not the thing itself.

What FTP is not is a fitness score. It is one point on a power-duration curve that runs from a standing-start sprint out to an all-day ride, and a single point cannot describe a curve. Two riders can test at exactly the same FTP and be completely different athletes. One has a huge sprint and a large anaerobic capacity, and comes apart once the effort stretches much past an hour. The other has almost no sprint and is still riding strongly deep into a long day. On a Zwift crit those two have almost nothing in common, and their identical FTP tells you nothing about which one wins. This is also why the zone table below matters more than the single headline number: the zones are what you actually train in, and the headline number only sets their scale.

FTP also is not a ranking. The Coggan power profile in the calculator above is an expert-derived reference scale, not a percentile of the riding population. Coggan built it by anchoring the extremes to known world-champion and untrained performance and interpolating between them, which means the bands in the middle are informed judgement rather than measurement. That is a perfectly reasonable way to build a coaching tool, and it is a bad way to work out whether you are "above average".

Finally, FTP is not a fixed property of your body that holds still. It moves with heat, with sleep, with how much training load you are carrying, and with the time of day. A number measured on the last day of a recovery week and the same number measured on the fourth day of a hard block are not the same measurement, even if the same rider produced both.

The 20-minute test

FTP is 95% of your average power for a single maximal, self-paced 20-minute effort. The 5% haircut removes the anaerobic contribution that lets a 20-minute effort sit above true hour power. It is the most widely used test in cycling, and its weakness is not physiological - it is that the result is only ever as good as your pacing.

Start too hot and you fade in the closing minutes, and the average comes out lower than if you had ridden the whole thing flat. Start too conservatively and you finish with something left, which produces a genuinely low number rather than a wrong one. Either way, the test punishes inexperience, which is exactly the group most likely to be doing their first test. If you have never held a hard effort for 20 minutes, do a ramp test first and use it to learn what your target should be.

The 8-minute test

Two maximal 8-minute efforts separated by 10 minutes of easy spinning, then FTP is 90% of the 8-minute power. This one came from CTS and Chris Carmichael, not from Coggan, which matters because it is routinely miscredited. The deeper 10% haircut exists because eight minutes carries a larger anaerobic contribution than twenty does.

There is an implementation split worth knowing about. CTS multiplies the higher of the two 8-minute averages by 0.90. Most apps average the two efforts first and then multiply. If your two efforts were 320 W and 300 W, the CTS method gives 288 W and the averaging method gives 279 W. That 9 W gap is produced by bookkeeping, not by physiology, and it is the kind of margin riders spend a whole training block chasing. Pick one convention, write it down, and never mix them.

The ramp test

Start at 100 W, add 20 W every minute, ride until you cannot turn the pedals, then take 75% of your best 1-minute power. The 75% approximates the ratio between threshold power and maximal aerobic power. It is the cheapest test on this page in every sense: it needs no pacing skill, it is the shortest of the lot, and it is extremely repeatable because there is nothing to get wrong. How long it lasts is set by the ramp itself, not by you - each step adds 20 W, so a stronger rider simply climbs more steps before failing.

Its failure mode is systematic and worth taking seriously. A rider with a big anaerobic capacity can keep grinding through two or three extra steps on anaerobic contribution alone, which lifts their best minute and hands them an FTP they cannot hold for an hour. Sprinters and track riders consistently get flattered by the ramp. In the other direction, a diesel with a small anaerobic capacity fails the ramp early despite a genuinely high threshold and gets under-rated. If your ramp result gives you zones where Z3 tempo work feels like the Z5 intervals in the table above, that is the mechanism, and the fix is a longer test rather than more willpower.

The 60-minute test and the 20 km TT

A 60-minute test or a ~40 km time trial has a multiplier of 1.00 for the simple reason that it is not an estimate at all. This is Coggan's primary definition, and everything else is derived from it. It costs the most, it is the hardest to schedule, and almost nobody repeats it often enough to track a training block. But there is nothing to argue about afterwards.

Coggan also published a 0.93 factor for a 20 km time trial, half the distance of the ~40 km effort he brackets at 50 to 70 minutes. It is the most convenient option here if you already ride a fixed 20 km course or Zwift route, because the effort exists in your history whether or not you called it a test.

The Joe Friel 30-minute test

Ride 30 minutes solo as hard as you can hold, then take the average power of the final 20 minutes with no correction factor at all. Friel's reasoning is that a solo test is already about 5% self-limited by the absence of race motivation, so the haircut is built into the situation rather than applied afterwards. His pacing instruction is the most useful sentence in any test protocol: the first 10 minutes should feel almost too easy. Because only the last 20 minutes count, a restrained opening costs you nothing and protects the part that does.

Any effort shorter than an hour includes an anaerobic contribution that you could not sustain for an hour. The shorter the test, the bigger that contribution, and the bigger the haircut needed to strip it out. That is the whole logic of the ladder: 1.00 at an hour, 0.95 at twenty minutes, 0.90 at eight minutes, 0.75 off a one-minute maximum in a ramp.

What those numbers are not is physiological constants. They are population averages applied to an individual, and the size of your own anaerobic contribution is exactly the thing they cannot see. The best evidence that even the author treated them as rules of thumb is the 20 km factor: Coggan published 0.93 for a 20 km time trial. Twenty kilometres is a distance, not a duration. Coggan's own bracket for the ~40 km version is 50 to 70 minutes, which is a 40% spread in duration for one fixed distance, and the same spread exists at 20 km. So one flat correction factor is being applied to a strong rider's effort and to a much longer one from a slower rider, even though duration is precisely the variable the factor is supposed to be correcting for. Nobody derives a physiological constant that way. You derive a useful approximation that way, publish it, and expect coaches to sanity-check it against the athlete in front of them.

The practical consequence is that switching protocols mid-season is worse than useless. If you ramp-test in January and 20-minute test in March, the difference between the two numbers is mostly protocol, not fitness. The reverse table in the calculator shows how large that effect is: hold the target FTP and the weight fixed, switch the protocol, and the raw number you have to produce moves by more than a season of honest training would move it.

Zwift ships two families of test workout. The FTP Test and the shorter version both take 95% of a 20-minute test interval, so they are the Allen and Coggan protocol with a guided warm-up attached. The Ramp Test starts at 100 W, adds 20 W a minute in ERG mode, asks you to stay seated, and takes 75% of your highest 1-minute average. Ramp Test Lite uses the same 0.75 calculation but starts at 50 W and steps up 10 W a minute, which is designed for riders under about 60 kg or with an FTP below roughly 175 W. If you are a lighter rider and the standard ramp feels like it starts halfway up your range, that is because it does.

The Grade is different again. It is a 3.5 km segment averaging 8.6%, launched with the June 2024 Watopia expansion, and instead of one fixed multiplier Zwift applies a duration-dependent curve running from roughly 85% to 100%, built from over 700,000 historical Zwift FTP tests. At a 20-minute completion time the multiplier lands at about 0.95, which is why it is not offered as a fixed option in the calculator above: your multiplier depends on how long you took, so a single dropdown entry would be misleading. If you tested on The Grade, take the number Zwift gave you.

zFTP is not a test at all. Since December 2024 Zwift has detected your threshold passively, modelling it from your power-duration curve using efforts of roughly 8 to 60 minutes during normal riding. There is no fixed multiplier to reproduce, which is why this calculator does not offer one.

That also explains why zFTP disagrees with your test result. A tested FTP is one controlled effort on one day. zFTP is a curve fitted to whatever your recent hard riding happened to contain. If your last month included a race where you sat above threshold for twelve minutes on a wheel, the model has better data than your test did and zFTP goes up. If you have spent six weeks doing endurance rides and 30-second sprints, the model has nothing in the 8 to 60 minute window to fit and it will drift or lag. Neither number is lying. They are answering slightly different questions, and the tested one is the one you should set your workout zones from.

Most bad FTP numbers are bad executions, not bad protocols. This is the sequence that produces a number you can trust.

1. Arrive fresh. Come off an easy day, ideally the last day of a recovery week. Testing deep inside a hard block measures your fatigue, not your fitness, and you will then train off that fatigue for the next two months.
2. Set the room up first. Fan pointed at your chest and face, room as cool as you can get it, bottle within reach, phone on silent. Heat is the single largest controllable variable in an indoor test - see the section below for the numbers.
3. Warm up properly, and longer than feels necessary. Build from easy spinning in Z1 up into Z2 and then briefly into Z3, add a few short bursts at high cadence with easy spinning between them, then drop back to Z1 until your breathing has settled and your legs feel clear rather than heavy. A cold rider produces a low number, and the penalty for under-warming is far larger than the cost of ten extra easy minutes.
4. Turn ERG mode off. ERG holds a fixed wattage and fights you when you fade or surge. A test is a self-paced effort, so you need slope or level mode and control of your own power.
5. Pick your opening target before you start. The test is arithmetically self-defining: to reproduce a previous FTP you need a 20-minute average of that FTP divided by 0.95, so open there and not above it. If you have no previous number, do a ramp test first and run its result through the calculator to get the target.
6. Minutes 0-5: it should feel sustainable. Hard, but clearly something you could keep doing. If you are already grimacing at minute three you have opened above your real ceiling, and the correct response is to back the power off immediately rather than to hope. A corrected pace still gives you a usable number; a blow-up does not.
7. Minutes 5-15: hold it and stop looking at the number. This is the part that decides the test. Chasing a dip with a surge costs more than the dip did.
8. Minutes 15-20: spend everything left. Lift from minute 17 if you can, and empty yourself in the last 60 seconds. If you could have gone harder in the final minute, the number is honest but conservative.
9. Take average power for the interval. Not normalised power, which inflates any effort that was not perfectly steady. Then multiply by 0.95, or put the raw number into the calculator above.
10. Use the same power source every time. A trainer and a power meter routinely disagree by a few percent. That offset is stable and harmless as long as you never switch mid-season.

The honest framing is this: FTP is a physiological property and it does not change when you move from the road to the pain cave. What changes is your ability to express it. The dominant mechanism is heat. Outdoors you ride through moving air that strips sweat off your skin. Indoors you sit still in your own microclimate, core temperature climbs, blood is diverted to the skin for cooling, and the same physiology produces fewer watts at the same perceived effort.

Two studies bound the problem usefully. Mieras, Heesch and Slivka (2014) had twelve recreationally trained male cyclists ride 40 km outdoors and 40 km in the laboratory at the same perceived effort. Outdoor power averaged 208.1 ± 10.2 W against 163.4 ± 11.8 W in the lab, a highly significant gap, with higher skin temperature indoors (33.0 against 31.4 degrees C). The caveat matters more than the headline: the indoor trial used no fan. That roughly 30% gap is what happens in still air, and it is not what a well-cooled Zwift setup produces. Do not quote it as your Zwift-versus-outdoor number.

Peiffer and Abbiss (2011) isolated the mechanism directly. Nine trained male cyclists rode four 40 km time trials at 17, 22, 27 and 32 degrees C. Mean power was 309 ± 35 W at 32 C against 329 ± 31 W at 17 C: about 20 W, roughly 6%, lost across a 15 degree rise. Very roughly, that is 1.3 W per degree C.

Read those two together and the practical conclusion is unavoidable. A big fan is the highest-value purchase in a pain cave, ahead of a better trainer, ahead of a smart bike, ahead of anything aerodynamic. At roughly 1.3 W per degree C, pulling ten degrees out of your effective riding temperature is worth on the order of 13 W, which is more than most riders get from an equipment upgrade costing many times as much. Nothing else in the room converts money into sustainable watts at that rate.

For riders who train indoors regularly and are well cooled, the gap commonly quoted is 3-8%, or 20-30 W in absolute terms. Treat that as a commonly cited range rather than a measured constant - it is a summary of coaching experience, not a controlled result, and your own figure depends almost entirely on your cooling.

The operational advice follows from the framing. Test where you train and race. If your racing happens on Zwift, set your Zwift zones from an indoor test, because an outdoor number will prescribe intervals you cannot complete indoors. If you do both, keep two numbers and stop treating the difference as a mystery. Zwift versus outdoor cycling goes further into what else changes between the two.

Once per training block is plenty, placed at the boundary and ideally on the last day of a recovery week. Testing more often than that spends training days on measuring rather than improving, and a maximal test costs you the session it replaces plus part of the next one. A couple of well-placed tests across a season will tell you everything a monthly test would, with far less disruption and far less chance of catching yourself on a bad day and believing it.

Between tests, your workouts tell you whether the setting is right. Read the signals rather than waiting for a calendar date.

You do not need a full retest to fix either case. Nudge the setting by a few watts in the indicated direction and run the next block off it. That is a legitimate way to converge on a working number, and unlike a retest it costs you no training time at all. A formal test is for establishing a baseline you will compare against months later, not for maintaining week-to-week zones. The training plans guide and the periodisation article both cover where in a plan a test belongs.

The Coggan chart is a reference scale. If you want an actual distribution, Cycling Analytics published one in June 2018 from thousands of power-meter users, using each rider's best 20-minute power as an FTP proxy. These are men's figures in raw watts, not W/kg, so they say nothing about how fast anyone climbs.

Two things are worth noticing. The median is 286 W of 20-minute power, which is about 272 W of FTP after the 0.95 haircut. Carried by a 75 kg rider that is roughly 3.6 W/kg, which lands in the Good band on the men's chart rather than anywhere near Untrained - a reminder that people who buy power meters are not a random sample of humans, and that comparing yourself to this table is comparing yourself to the committed. And the spread from the 25th to the 90th percentile is only about 100 W. Most riders are far closer together than the internet suggests, which is worth remembering the next time a forum thread makes the 90th percentile sound like an entry requirement.

On a steep climb almost all of your power goes into lifting your own mass, so climbing is settled by watts per kilogram rather than watts. This is where the second number in the calculator earns its place. Somewhere around 3.2 to 3.3 W/kg is the hour mark on Alpe du Zwift. At 75 kg that is about 240 to 248 W, and at 60 kg it is about 192 to 198 W. Nearly 50 W separates those two riders and they arrive within a couple of minutes of each other, which is the whole argument for tracking W/kg alongside watts.

That also means the same FTP produces completely different climbing outcomes depending on what you weigh, which is why every climb tool on this site takes weight as an input. The Alpe du Zwift calculator is fitted to real finishing times on that one climb, while the climb time calculator runs the physics from first principles and works on any sustained climb in the game. If you want to see the full list of what is climbable and how long each one is, the Zwift climbs index has the numbers, and Alpe versus Ven-Top compares the two big ones directly.

One caution. Because W/kg is a ratio, it moves when either number moves, and the arithmetic makes weight loss look like an easy route to a better climbing figure. It is not free. Losing mass usually costs some power, the ratio improves less than the arithmetic suggests, and under-fuelling a training block is the fastest way to lose the FTP you just measured. The calculators here treat your weight as an input, not a recommendation.

This page is training information, not medical advice. A maximal FTP test is a genuinely hard effort - a ramp test ends in voluntary failure by design, and a 20-minute test is one of the harder things you will do on a bike. If you have a heart condition, high blood pressure, are returning from illness, are taking medication that affects heart rate, or you are new to structured exercise, talk to a doctor before doing one. If you get chest pain, dizziness or an irregular heartbeat during a test, stop.

Every number on this page comes from a published source. The protocols and their multipliers, the Coggan zone table, the power profile chart, the two heat studies and the Cycling Analytics distribution are all listed below so you can read the originals rather than taking this page's word for it. Corrections are welcome and get applied - send them here, and the editorial policy explains how they are handled. This calculator is built and maintained by Christian Lassen Dam.