Zwift FTP Test: Ramp, 20-min, The Grade & zFTP Accuracy (2026)
Zwift gives you three ways to test your FTP — Ramp, 20-min, and The Grade — plus a fourth number it maintains automatically, zFTP. Each reads your engine a little differently, and the right one depends on your physiology.
Why FTP Matters Inside Zwift
FTP is the wattage you can hold for about an hour. Get the number wrong, and every Zwift workout in ERG mode runs at the wrong intensity. Three points wrong on FTP means 9–12 W mis-targeted in every Zone 4 interval for the rest of the block — too easy and you under-train, too hard and you over-cook recovery.
A note on language: in this article "FTP" means the Zwift convention — your sustainable hour-power. In our coaching engine we call the same underlying value TP (Threshold Power, derived from a Critical-Power model rather than a 60-minute test). The numbers are usually within 3–8 W of each other; the methodology is what differs.
The Four Zwift FTP Protocols
| Protocol | Duration | Format | Evidence vs Mader TP | Best-fit rider type |
|---|---|---|---|---|
| Ramp | about 20 min | +20 W/min step ramp until failure | own cohort (1,818 tests): factor 67–83% | most riders, esp. VO2-dominant |
| 20-min | 45 min | warm-up → 5-min opener → recover → 20-min all-out | not analysed in this cohort | balanced / VO2-dominant |
| The Grade | >8 min (typically 15–20) | ride the Epic KOM climb at max steady effort | not analysed in this cohort | riders who prefer a climb |
| zFTP | automatic | Zwift estimates from your 90-day power curve | not analysed in this cohort | anyone racing recently |
Only the Ramp band is backed by our own data. It comes from 1,818 cycling ramp tests from 1,014 athletes, selected from over 15,000 Powertests on the platform by a deliberately strict standard: between the 5th and 95th percentile the real conversion factor runs from 67% to 83%. Across the whole cohort, though, Zwift's fixed 0.75 is well chosen — the mean and median land at 0.753 and 0.755. The full derivation is in Zwift FTP Test vs Powertest.
The other three rows are coaching experience, not a cohort analysis — and this is a deliberately selected group of athletes who train structured and use data feedback, not the population of all Zwift riders. Read the non-Ramp entries as orientation only. As of August 2026.
Protocol 1 — The Ramp Test
A 20-watt-per-minute step ramp until your legs give up. Easy at the start (you ride 80 W for the first minute), brutal at the end (most riders fail somewhere between 320–420 W). FTP is then calculated as 75% of the highest one-minute power you sustained.
Who wins. Zwift calibrated the average well: across our cohort the true conversion factor's mean and median are 0.753 and 0.755, within a watt and a half of the fixed 0.75. The catch isn't the average — it's that your personal factor varies with your physiology, and the direction of the miss flips with rider type.
Who loses — in both directions. Low-VLamax diesels came out around 0.83 in our data, so the fixed 0.75 tends to under-read their threshold. High-VLamax sprinters came out around 0.70, so it over-reads theirs — sometimes enough that the ERG workouts afterwards feel impossible. How many watts either miss costs depends on your ramp power.
Fatigue cost. Low. You can usually ride hard the next day.
Protocol 2 — The 20-Minute Test
A 5-minute all-out effort to open the legs, a 10-minute recovery, then 20 minutes at the highest power you can sustain. FTP equals 95% of that 20-minute average.
Who wins. Balanced riders and VO2-dominant cyclists who pace well. Twenty minutes is long enough that you cannot fake the number with anaerobic capacity — your aerobic system is the bottleneck.
Who loses. High-VLamax sprinters in the opposite direction — the 5-minute opener cooks them and they bleed power in the 20-minute block. Also anyone who has never paced a 20-minute effort; expect 5–10% pacing error on a first attempt.
Fatigue cost. High. Plan an easy day before and a recovery day after.
Protocol 3 — The Grade
Zwift's on-road FTP test, launched in June 2024. Instead of a structured workout, you ride a real climb — the 8–9% Grade up to the Epic KOM in Watopia — at the hardest steady effort you can hold. Zwift reads your average power over the climb and, with a model built from 700,000+ prior FTP tests, turns any completion over 8 minutes into an FTP estimate. Beat your stored number and Zwift updates your saved FTP automatically — it only ever revises upward, never down.
Who wins. Riders who'd rather race a climb than pace a test workout. Most complete it in 15–20 minutes.
Who loses. We haven't tested The Grade against our own cohort, so treat this as coaching judgment: because it rewards a big steady effort, a large anaerobic engine may still flatter the estimate, especially on shorter completions. If your Sweet-Spot work later feels too hard, cross-check with a 20-minute test.
Fatigue cost. Moderate — coaching judgment, and it varies with how long your climb takes.
Protocol 4 — zFTP
Alongside the FTP you set yourself, Zwift keeps a second, automatic estimate called zFTP — built from your best power over the last 90 days, with efforts between 12 and 50 minutes carrying the most weight. You can't edit it by hand: it tracks whatever you've recently done on the platform.
The two numbers do different jobs. Your set FTP scales your workouts and ERG targets. Your zFTP is what Zwift uses to suggest your race category. They're calculated differently and won't always agree — and changing one doesn't change the other.
Who wins. Anyone racing or doing hard group rides on Zwift regularly — the estimate is free and updates automatically from your recent activities.
Who loses. If your last 90 days hold no representative hard efforts — especially between 12 and 50 minutes — the estimate may not reflect your current sustainable power.
Worth knowing. Because zFTP updates itself from your recent rides, the category Zwift suggests can move even in a block where you never sat down to test — it reflects your recent Zwift efforts, not a controlled test. Zwift also refines how these numbers are computed over time, so treat the specifics here as the picture as of August 2026.
How Rider Type Maps to the Right Protocol
If you have ever done an A Faster You Powertest, you already know your VO2max/VLamax balance. The map below tells you which Zwift test to lean toward.
| Your engine | Lean toward | Be cautious with |
|---|---|---|
| High VO2max, low VLamax (diesel) | Ramp · zFTP | The Grade — may read a touch low |
| High VLamax, modest VO2max (sprinter) | 20-min · The Grade | Ramp — tends to over-read you |
| Balanced (allrounder) | The Grade · 20-min | none outright bad |
| Untested rider | The Grade | — |
The single most common mismatch is a high-VLamax rider taking a Ramp test, ending up with an FTP too high, and then struggling in the first Zone 4 ERG workout of the new block. If your Sweet-Spot intervals feel harder than they should, your protocol is the first place to look.
A Note on the Zwift–A Faster You Integration
Direct Zwift integration is live as part of the A Faster You plan. Your scheduled workouts sync directly to Zwift with the right ERG power targets, and your Zwift rides flow back into your plan as completed sessions — no more FIT-file exports through the upload page. The setup option lives at Settings → Connections → Zwift.
The plan still uses your tested FTP, not Zwift's auto-derived zFTP — so the choice you make in this article carries straight into your sync.
When You're Ready for the Real Number
All four Zwift protocols are useful when you do not have access to anything else. None of them measure what actually matters for training prescription: your underlying VO2max plus your VLamax, which together set your true threshold and your individualised training zones.
A Mader-model Powertest — a 10-second sprint plus a 12-minute all-out, recorded on the trainer or outdoors with one-second data — takes the same time as a Ramp test and returns both numbers, plus the zone boundaries that Zwift cannot compute. How close that VO2max lands to the lab is measured: mean bias −1.8 ml/min/kg across 43 athletes with parallel spirometry (the full validation). And to see where that number puts you, the VO2max table by age and gender ranks it against 15,000+ Powertests. Most of our athletes do their first one in week one of the plan and re-test every 8–12 weeks after that.
If you want to see exactly how much the protocol you picked above costs you in watts versus a Powertest, the next read in this series breaks it down with cohort data: Zwift FTP Test vs Powertest: When 300W Aren't 300W →
Start a free trial of A Faster You and your first Powertest sets the baseline. The AI schedules every interval, threshold session, and recovery day after that.
FAQ
Can I run two Zwift FTP tests in the same week to compare? You can, but the second result will read 3–5% low because of residual fatigue from the first. Space them at least 5 days apart, with two easy days in between.
What does Zwift do if my zFTP is higher than my set FTP? They do different jobs, so both can be true at once. Your set FTP drives your workouts and ERG targets. zFTP is a separate automatic metric — you can't edit it by hand — that Zwift uses to suggest your race category. Current race placement runs on Zwift Racing Score, seeded from your best 5-minute power over the last 90 days and your weight, not from an FTP-test result.
Should I do the Ramp test if I have a heart-rate monitor but no power meter? No — none of the four protocols work without power. Use a Zwift-compatible smart trainer or a power-meter pedal. Heart-rate-based FTP estimation is too noisy to set ERG targets from.
Does The Grade work for triathletes or only cyclists? The protocol is identical for both. The interpretation shifts slightly: triathletes often have a lower VLamax, which can make The Grade read a little low for them. If you race triathlon, the Powertest is a much better baseline than any of the Zwift options.
My Ramp test gave 280 W but my Sweet-Spot intervals at 235 W feel hard. What went wrong? You are likely a VLamax-dominant rider whose Ramp result is too high. Repeat the test using The Grade or a 20-minute protocol. If the result drops meaningfully, trust the lower number and re-build your zones from it.
How often should I re-test in Zwift? Every 8–12 weeks during a focused training block, or whenever your prescribed workouts start feeling consistently too easy or too hard. If you are on an A Faster You plan, the system flags the right moment for you.
Sources: Mader, A. (2003) — Glycolysis and oxidative phosphorylation as a function of cytosolic phosphorylation state and power output of the muscle cell. Eur J Appl Physiol, 88(4–5), 317–338. Mader, A. & Heck, H. (1986) — A theory of the metabolic origin of the "anaerobic threshold". Int J Sports Med, 7 Suppl 1, 45–65. Buchheit, M. & Laursen, P.B. (2013) — High-intensity interval training, solutions to the programming puzzle. Sports Medicine, 43. Zwift help docs as of 2026-05. Cohort behind the ramp band: 1,818 cycling ramp tests from 1,014 athletes, selected from over 15,000 Powertests on the A Faster You platform by the standard described above. As of August 2026.