Should I set my zones from a 20-minute FTP test?

A 20-minute test is a useful check. It is a bad master.

FTP is an estimate of the highest power you can hold for about an hour. The common field version is: ride 20 minutes as hard as you can, take the average, multiply by 0.95. That 5% cut is a population average, not your physiology.

For some riders it is close. For others it sets every zone 10–20 watts wrong.

The problem

One number cannot describe a curve.

A rider with a big anaerobic reserve can fake a high 20-minute average and still fall apart at minute 35. A diesel can look “low FTP” on a 20 and then hold that power far longer than the test predicted.

Then the plan does this:

• Z2 is set off the inflated number, so “easy” rides sit too hard
• Threshold intervals miss the actual heavy-to-severe boundary
• VO2 targets are either junk or heroic
• You chase a test you can game instead of the durations your event actually uses

The test is reliable when you repeat the same protocol. Reliable is not the same as valid for zones.

The rule

Do not let a single 20-minute file set the whole week.

Use three anchors from real maximal or near-maximal efforts — tests or races, not guesses:

• 5 minutes — aerobic ceiling / MAP region
• 20 minutes — the usual field check
• 40–60 minutes — the number that should actually behave like threshold

If you only have two, use 5 and 20. If a recent race gave you a clean 40–60 minute power, trust that more than a fresh 20-minute test on a bad day.

Then set work from the duration you are training:

• Endurance work from the long number, not from 95% of a 20
• Threshold work near the power you can actually stabilize, not the one you spiked for 20 minutes
• VO2 work from the 5-minute number, not “120% of FTP”

A power-duration curve beats a single FTP. Critical power from several maximal efforts (roughly 3–20 minutes) is a better boundary than 20 × 0.95 for most trained riders — and it still is not a license to live at that line all week.

How the 0.95 trick fails

The 0.95 factor assumes a typical anaerobic contribution to a 20-minute effort.

It overestimates threshold when:

• you are punchy (crit, track, strong 1–5 min)
• you started the 20 too hard and the average is a fade, not a hold
• the test is indoors with extra cooling and no steering load

It underestimates threshold when:

• you are a time-trial or gran-fondo diesel
• you paced the 20 cautiously
• you already have a recent 40–60 minute race file that sits above 0.95 × 20-min

Studies that compare 20-minute FTP to laboratory markers (MLSS, CP) usually find a strong correlation and wide individual error. Correlation is why the test survives. The error bars are why your zones feel wrong.

What to do this month

Stop testing every four weeks “because the app said so.” A hard race or a well-paced 20 on fresh legs is enough every 6–8 weeks, or when the work stops matching the feel.

Pull best 5-min, 20-min, and longest steady effort from the last 6–8 weeks. Calibrated meter. Same bike if you can.

If 20-min × 0.95 is more than ~5% above a recent 40–60 min hold, drop the threshold target toward the long number.

If 5-min power is not at least ~15–20% above threshold, your “VO2” days are just more threshold. Fix the ceiling, don’t raise FTP on paper.

Keep Z3+ inside the 3-hour / 20% cap. A prettier FTP does not buy more hard minutes.

The last check

If threshold intervals always die at 8–10 minutes, the number is high. If they feel like tempo, the number is low. Believe the repeatable session, not the one test you padded.

Zones exist to hit a physiological target. A 20-minute average is one data point on the way there.