How to increase your FTP: what a real gain costs, and how to tell one from noise

What counts as a real FTP gain (and why most reported gains are not)

A gain is real when it exceeds the measurement error of the test that found it. For the 20-minute protocol that error sits between 1.3% and 2.9%, depending on how well the test is controlled [MacInnis et al. 2019, Borszcz et al. 2020]. On a 250-watt FTP, a provable change starts between 9 and 20 watts.

Two studies bracket the number. Borszcz and colleagues [Borszcz et al. 2020] had 25 trained cyclists repeat the full 20-minute protocol seven days apart and reported a coefficient of variation of 2.9% with an intraclass correlation of 0.97 — reliable by the standards of sports science, and still loose enough to matter. MacInnis and colleagues [MacInnis et al. 2019] found a typical error of 4.6 watts on 20-minute trials averaging 343 watts in eight highly trained riders, about 1.3%. The gap between them is not a contradiction. Eight well-drilled riders doing controlled laboratory trials is the best case; 25 riders a week apart is closer to your situation.

Reliability converts into a per-rider threshold through a standard piece of arithmetic: the smallest change one individual can detect at 95% confidence is roughly 2.8 times the typical error. At the 2.9% end, a 250-watt FTP has to move about 20 watts before one re-test separates the gain from the noise. At the 1.3% end it is about 9 watts. Every jump smaller than that — the +6 after a hard block, the +8 after a new position — is a number you cannot defend from a single test, however good the block felt.

Two further sources of error sit on top of that. Tramontin and colleagues [Tramontin et al. 2022] showed that the warm-up preceding a 20-minute test shifts the resulting FTP on its own, and Borszcz's data put the coefficient of variation of the 45-minute warm-up at 5.5%, nearly twice that of the test it precedes [Borszcz et al. 2020]. The 20-minute protocol is also an estimate of something else: against a real 60-minute effort, limits of agreement ran from -40 to 32 watts in one comparison and -62 to 60 watts in another [Borszcz et al. 2018]. Keep two questions separate here. Whether a change is detectable is a measurement question. Whether it matters is a racing question, and a 1% change can decide a race while staying invisible in a test. The common mistake is answering the second question with the first.

How much amateurs actually gain, and how long it takes

Controlled trials in trained cyclists move threshold-region power by roughly 2% to 5% over three to twelve weeks of focused work [Rønnestad et al. 2020, Rønnestad et al. 2014]. Riders with less training history gain more. The group averages hide large individual spread, and that spread is the part that applies to you.

The most directly relevant number comes from Rønnestad and colleagues [Rønnestad et al. 2020]. Nine elite cyclists did three weeks of short intervals — nine sessions in total — and improved mean power in a 20-minute test by 4.7%. The effort-matched long-interval group went backwards by 1.4%. Read the spread before the mean: the short-interval improvement was 4.7 ± 4.4%, so riders inside the same protocol ranged from roughly no change to about 9%. Nine riders per group is a small sample, and these were athletes at 73 mL/kg/min, not the amateur riding six hours a week.

Longer blocks produce larger numbers and thinner certainty. Over 12 weeks, block periodization raised power at 2 mmol/L lactate by 22 ± 14% against 10 ± 7% for a traditional layout [Rønnestad et al. 2014] — a result quoted as settled in a lot of coaching content that did not actually reach significance in the paper (P = 0.054, with 8 riders against 7). The VO2max difference in the same study did reach it: 8.8% against 3.7%. In recreational cyclists, a 16-week pyramidal programme moved power at 4 mmol/L lactate with a large effect size across 14 riders [Magalhães et al. 2024], though that study had no comparator group, so it shows the riders improved rather than that the distribution caused it.

The practical translation: a trained amateur running a focused 8-to-12-week block should plan for something in the range of 3% to 8% on threshold-region power — roughly 8 to 20 watts on a 250-watt FTP — and should expect a genuine chance of measuring nothing. Whether riders differ in trainability or the differences are measurement artefact is still argued. When Hecksteden and colleagues applied several accepted analytical methods to the same trial, only 11 of 20 subjects were classified the same way by all of them [Hecksteden et al. 2018]. That work used untrained subjects and VO2max rather than cyclists and FTP, so treat it as a caution about responder labels rather than a cycling result. It is enough to make 'I must be a non-responder' a claim nobody can currently support.

What raises FTP: the levers ranked by how well the evidence supports them

Four levers have controlled evidence behind them in trained cyclists: the format of hard intervals, how those sessions are arranged across weeks, heavy strength training, and consistent aerobic volume. Ranked by evidence quality rather than by how often they get recommended, interval format and strength training are the best documented.

Interval format. Rønnestad's group has tested short intervals — 30 seconds hard, 15 seconds easy, in long series — against effort-matched long intervals twice. Over 10 weeks in competitive cyclists, short intervals produced a larger VO2max gain than long intervals, 8.7% against 2.6%, with moderate-to-large effect sizes favouring short intervals across the measured parameters [Rønnestad et al. 2015]. Over three weeks in elite riders, the same comparison moved 20-minute power by 4.7% against -1.4% [Rønnestad et al. 2020]. Both studies are small and both come from one research group, which is the honest caveat. The mechanism is not mysterious: short recoveries stop oxygen uptake falling between efforts, so a set accumulates more time near VO2max than a longer interval at the same perceived effort.

How the hard sessions are arranged. Block periodization concentrates five high-intensity sessions into one week, then drops to one per week for three. Over 12 weeks that produced a larger VO2max improvement than two sessions every week, 8.8% against 3.7% [Rønnestad et al. 2014]. The broader question of intensity distribution has weaker support than its popularity suggests: Rosenblat and colleagues found only four randomised trials comparing polarized against threshold-model training that met their inclusion criteria, three of which entered the meta-analysis, and reported a moderate pooled effect favouring polarized [Rosenblat et al. 2019]. A moderate effect from three trials of moderate methodological quality is a reason to lean polarized, not a reason to be certain.

Heavy strength training. Nineteen well-trained female cyclists were randomised to 11 weeks of endurance training with or without heavy leg strength work [Vikmoen et al. 2016]. The strength group improved mean power in a 40-minute all-out trial, cycling economy, and the fraction of VO2max they could sustain. The endurance-only group changed on none of those measures. This is the lever amateurs skip most reliably, and one of the few where a randomised trial in cyclists shows a clean group difference on a performance measure rather than on a laboratory index.

Aerobic volume. In the 16-week recreational-cyclist study, the improvement in power at 4 mmol/L lactate correlated most strongly with time spent in the moderate zone between 2 and 4 mmol/L, not with time above it [Magalhães et al. 2024]. A correlation inside a single-arm study is weak evidence, and it points the same way as everything else: the hard sessions take the credit, the easy hours make them repeatable. Getting the easy hours genuinely easy means knowing where your zones fall in watts, which is arithmetic on one number — a power zones calculator turns your FTP into the seven bands you actually ride to.

Watts or kilograms: the two ways to move W/kg, and what each costs

Watts per kilogram has a numerator and a denominator, and the two are not symmetric. For a 75-kilogram rider at 250 watts, one kilogram of body mass is worth about 3.3 watts. Matching a 20-watt FTP gain through weight alone means losing 5.6 kilograms.

Run the numbers. A 75-kilogram rider with a 250-watt FTP sits at 3.33 W/kg. Add 20 watts and they are at 3.60. Drop to 73 kilograms at unchanged power and they are at 3.42, about a third of the same gain. To reach 3.60 on weight alone, that rider has to get to 69.4 kilograms. The general shortcut is that one kilogram is worth roughly your current watts-per-kilogram figure converted into watts: 3.3 watts for a rider at 3.3 W/kg, 4.5 watts for a rider at 4.5. Riders at a higher W/kg get more from each kilogram, which is exactly the group with the least spare mass to give.

The denominator looks cheaper because the arithmetic quietly assumes the numerator holds still. Sustained low energy availability degrades endurance performance alongside bone, immune and endocrine function, and the 2023 International Olympic Committee consensus statement on Relative Energy Deficiency in Sport lists performance decrement among its defining outcomes [Mountjoy et al. 2023]. A rider who strips 5 kilograms across a build block by under-eating usually arrives lighter and weaker, with W/kg unchanged. The 5.6-kilogram calculation above holds only if the watts survive, and aggressive deficits are the condition under which they tend not to.

Which lever is right depends on where you are. A rider carrying genuine excess mass has the cheaper path in the denominator and should take it slowly, outside a build block. A rider already near a sustainable racing weight has only the numerator. And W/kg decides climbing while deciding nothing about a flat criterium, where absolute watts win. A power profile that places your watts per kilogram against a population at 5 seconds, 1 minute, 5 minutes and 20 minutes will usually tell you faster than any chart whether your problem is the numerator, the denominator, or a duration you have been ignoring.

When the number stops moving: what to change, in what order

Check in this order: the test, then recovery, then the easy rides, then the hard ones. Most self-coached riders reverse it and add intensity first, which is the change most likely to make a plateau worse. Three of the four fixes involve doing less rather than more.

Test the test first. Before concluding that training stopped working, confirm the measurement did not change. Same warm-up, same trainer or same climb, same time of day, same fuelling, same week of the block. The first section put numbers on why: the warm-up alone varies nearly twice as much as the test it precedes, and its structure shifts the result on its own [Borszcz et al. 2020, Tramontin et al. 2022]. A plateau measured under two different protocols is not a plateau. It is two measurements.

Then recovery. The block-periodization result is not only about concentrating intensity; it is equally about the three weeks of one hard session that follow each loading week [Rønnestad et al. 2014]. A rider doing two or three hard sessions every week for 14 straight weeks is running the arm of that trial that improved less. Then the easy rides. If your endurance rides sit at upper tempo, you are neither accumulating the aerobic hours that tracked threshold improvement in recreational riders [Magalhães et al. 2024] nor arriving fresh enough for the hard sessions to be hard.

Only then change the hard sessions. Swap long intervals for a series of 30/15s, or concentrate five hard sessions into one week and back off for three [Rønnestad et al. 2020, Rønnestad et al. 2014]. Add heavy leg strength twice a week if it is missing [Vikmoen et al. 2016] — the lever with the cleanest randomised evidence in cyclists and the one most often absent from an amateur week. Give any of these 8 to 12 weeks before judging them. Three weeks was the shortest interval over which any trial cited here detected a change, and that was in elite riders doing nine sessions.

Why your FTP can rise while you get slower

FTP is one coordinate on a power-duration curve, and a block that lifts that coordinate can flatten everything around it. A rider who spends a winter on threshold work often gains 15 watts of FTP and loses the sprint, the 1-minute power, and the ability to produce anything after three hours.

The short-interval protocol moved both ends of the curve: Rønnestad's group reported superior adaptations in the high-power and the lower-power regions of the profile against effort-matched long intervals [Rønnestad et al. 2015]. That was the design of one protocol, not a property of training in general. Spend twelve weeks exclusively at threshold and the 20-minute point usually rises while the 1-minute and 5-second points fall, because nothing trained them. For a road racer whose events are decided by a 30-second gap, that trade is a net loss wearing a better headline number.

There is a second way the headline misleads. Borszcz and colleagues measured how long trained cyclists could actually hold the power their 20-minute test called FTP: 50.9 minutes on average, with a standard deviation of 15.7 minutes [Borszcz et al. 2018]. Some riders held it past the hour; others failed well inside 40 minutes. Two riders with identical FTPs can therefore own very different amounts of usable time at that power, and a block that raises the number without raising the time you can hold it has not made you faster at anything you do on a Saturday.

This is why we grade four durations rather than one. A profile that places 5-second, 1-minute, 5-minute and 20-minute power separately against a population shows which coordinate moved and which one paid for it. Most riders find their shape is already lopsided, and that the duration they keep training is the one they were already best at. The useful question is rarely 'how do I raise my FTP' in isolation. It is which part of the curve is costing you the result you actually want.

Quick answers

Sources cited in this guide

1. 01
   Borszcz et al. 2020. Reliability of the Functional Threshold Power in Competitive Cyclists. International Journal of Sports Medicine.
2. 02
   MacInnis et al. 2019. The Reliability of 4-Minute and 20-Minute Time Trials and Their Relationships to Functional Threshold Power in Trained Cyclists. International Journal of Sports Physiology and Performance.
3. 03
   Borszcz et al. 2018. Functional Threshold Power in Cyclists: Validity of the Concept and Physiological Responses. International Journal of Sports Medicine.
4. 04
   Tramontin et al. 2022. Functional Threshold Power Estimated from a 20-minute Time-trial Test is Warm-up-dependent. International Journal of Sports Medicine.
5. 05
   Rønnestad et al. 2020. Superior performance improvements in elite cyclists following short-interval vs effort-matched long-interval training. Scandinavian Journal of Medicine & Science in Sports.
6. 06
   Rønnestad et al. 2015. Short intervals induce superior training adaptations compared with long intervals in cyclists - an effort-matched approach. Scandinavian Journal of Medicine & Science in Sports.
7. 07
   Rønnestad et al. 2014. Effects of 12 weeks of block periodization on performance and performance indices in well-trained cyclists. Scandinavian Journal of Medicine & Science in Sports.
8. 08
   Rosenblat et al. 2019. Polarized vs. Threshold Training Intensity Distribution on Endurance Sport Performance: A Systematic Review and Meta-Analysis of Randomized Controlled Trials. Journal of Strength and Conditioning Research.
9. 09
   Vikmoen et al. 2016. Strength training improves cycling performance, fractional utilization of VO2max and cycling economy in female cyclists. Scandinavian Journal of Medicine & Science in Sports.
10. 10
    Magalhães et al. 2024. Effects of a 16-Week Training Program with a Pyramidal Intensity Distribution on Recreational Male Cyclists. Sports (MDPI).
11. 11
    Hecksteden et al. 2018. Repeated testing for the assessment of individual response to exercise training. Journal of Applied Physiology.
12. 12
    Mountjoy et al. 2023. 2023 International Olympic Committee's (IOC) consensus statement on Relative Energy Deficiency in Sport (REDs). British Journal of Sports Medicine.

Specific questions inside this topic

Focused reads — each answers one question end-to-end.

1. 01

   How long does it take to raise your FTP by 20 watts?

   Twenty watts is near the smallest gain one 20-minute re-test can prove. What the trials observed, and why 20 W differs at 200 W and at 350 W.

   →
2. 02

   Watts or weight: which actually moves your W/kg?

   One kilogram is worth about your current W/kg in watts. Why that makes the denominator look cheaper than it is, and where W/kg is the wrong metric.

   →

Related guides

• FTP without a test: estimating threshold from real rides →
• The self-coached cyclist: training without a coach in 2026 →
• Adaptive cycling training plans that survive real life →