Why Your FTP Drops After Two Hours (and What to Do About It).
Written by Damian Ruse
Founder of SEMIPRO Cycling. Fourteen years coaching cyclists from beginners to the WorldTour.
The short version.
The number on your head unit was measured fresh, often after a rest day. Races and long rides are not fresh. As you accumulate work your threshold falls, your zones shift down with it, and the effort that felt like tempo at the start becomes threshold without you noticing. If you keep pacing off your fresh test, you cook yourself. The fix is to know your real late-ride threshold, train power under fatigue, and pace by feel and kilojoules, not just by the numbers from a rested test.
Your threshold is a snapshot, not a constant.
When you test your FTP, you capture what you can hold when rested. That number is real, but it is real for that moment. Once you are an hour or two into a hard ride, the conditions have changed. Glycogen is lower, the cost of every watt has risen, and the ceiling you tested has quietly come down.
The research puts the slide at around two percent an hour once the riding gets hard. Two percent does not sound like much, but it compounds. Over a three hour race that is a six percent drop. Over a long mountain stage it can reach ten percent. The number you planned your race around stopped being accurate well before the race was decided.
What this looks like in real race data.
Take two experienced masters racers at a recent hard mountain bike stage race, similar level, racing the same course on the same day. Call them Rider A and Rider B. From the race files, Rider A's critical power came out at 342 watts and Rider B's at 371 watts. Both strong. Both in the age range where durability decides the day.
Across Rider A's long stage, nearly six hours of racing, his critical power did not hold at 342. It fell to about 308 by the end, a ten percent drop in his ceiling. He was not just riding above threshold late in the stage. He was riding above a threshold that was shrinking every hour.
And his zones moved with it. The same effort meant something different by the back half of the race.
| Zone | Fresh | After the threshold declined |
|---|---|---|
| Zone 4 (threshold) starts at | 311 W | 280 W |
| Zone 3 (tempo) starts at | 260 W | 234 W |
An effort that felt like solid tempo at the start was now threshold. An effort that felt like threshold was now over his critical power. He was working in a higher zone than he thought, and paying for it without understanding why. His cadence told the same story, drifting from about 91 rpm down to 80 by the end as the legs turned slower under fatigue.
Same average power, fourteen minutes apart.
Here is the part that should change how you pace. Over the stage, Rider A averaged 302 watts and Rider B averaged 299 watts. Three watts apart. But Rider B finished nearly fourteen minutes faster.
| Rider A | Rider B | |
|---|---|---|
| Average power | 302 W | 299 W |
| How the watts were spent | Hard early, above CP on the climbs | Even, hour to hour |
| Result | Paid for it in the back half | Nearly 14 min faster |
Same engine on paper. Fourteen minutes apart in reality. That gap is not fitness. It is durability and pacing. The watts you spend in the first hour cost more than the watts you save for the third.
A coaching note on reading fatigue.
One thing worth knowing if you analyse your own files. When I first looked at Rider A's long stage using time as the x-axis, the standard decoupling calculation said he got more efficient over the stage, about minus three and a half percent. That is obviously wrong. You do not get more efficient over six hours of racing.
The problem was the terrain. The descents, the stops, the variable pacing distorted a time-based view. When I switched the x-axis to cumulative kilojoules, total work done, the real picture appeared: about plus four and a half percent drift across nearly five thousand kilojoules. Time can lie. Kilojoules do not. This is how we are starting to read fatigue at SEMIPRO, and it changes what you see.
What to do about it.
- Know your real threshold. The number from a fresh 20-minute test is your rested ceiling. Your working threshold after two or three hours is roughly ten percent lower. Plan your race effort around the number that will actually exist late, not the one you tested fresh.
- Train durability on purpose. Put quality work at the end of long rides, intervals after you are already tired, not only in the fresh first 90 minutes. Teach your body to make power when it does not want to.
- Pace by feel and kilojoules, not just zones. Your zones shift down as you tire. Watch how much total work you have done, not just how long you have been out, because two riders can ride three hours and accumulate very different fatigue.
- Fuel properly. Cardiac drift accelerates when glycogen runs low. Every gram of carbohydrate you skip early is threshold power you give away later. It is the cheapest durability gain there is, and most riders still get it wrong.
The verdict.
Your FTP dropping after two hours is physiology. You cannot switch it off. But how much it drops is trainable, and how well you pace around it is a skill. Test well on a Tuesday and you have a number. Race well on a Saturday and you have durability. The second one is what decides the day.
Common questions.
Why can I hold 280 watts fresh but barely 230 after two hours?
Because your threshold dropped underneath you. Power declines by roughly two percent an hour once a ride gets hard, so the ceiling you had fresh is simply lower late. It is not weakness or bad legs. It is durability, and it is trainable.
Does this mean my FTP test is wrong?
No. The test is an accurate snapshot of your rested ceiling. The mistake is using that one number for everything, including the back half of long races where your real threshold is well below it. The number is right. The way it gets used is the problem.
How much does FTP actually drop over a long ride?
Around two percent an hour once you are past the first 45 minutes. That is roughly six percent over a three hour race and up to ten percent over a long, hard stage. On a 300 watt threshold, that is a ceiling closer to 270 by the time it counts.
Someone with my numbers keeps dropping me late. Why?
Almost always durability and pacing, not raw fitness. Two riders can average nearly identical power and finish many minutes apart if one spends too much early and fades while the other rides even. The watts you save early are worth more than the watts you spend.
Want the full picture this comes from? Read about the durability framework, or see how a structured plan applies it.