Seven watts decided whether one age-grouper ran an Ironman marathon or walked most of it. That’s the average gap researchers found between Critical Power (CP) and FTP (Functional Threshold Power) in a group of trained cyclists, and outdoors, on race day, the real gap can run much wider.
Pace off the wrong number and you spend fuel on the bike you’ll need on the run.
Your FTP is the highest power you can hold for about an hour before you fall apart. Your Critical Power is a related idea measured a different way: instead of one long test, it comes from several short, hard efforts plotted on a curve, then stretched out to a theoretical “forever” power. Both claim to mark your threshold. They don’t always agree.
Critical Power vs FTP: The 7-Watt Gap the Research Found
A 2021 study in Frontiers in Physiology tested 17 trained cyclists and triathletes with both a Critical Power protocol and a standard 20-minute FTP test. Average CP came out at 256 watts. Average FTP came out at 249 watts. That’s a 7-watt gap, and the math gave a 91.7% chance CP really is the bigger number.
In plain terms: CP usually runs a touch higher than FTP. Not always by much, but often enough to matter.
Here’s the catch. The two numbers correlated tightly across the group (r = 0.969), but individual riders differed by anywhere from 19 watts lower to 33 watts higher than the average gap. Two athletes with “the same” 250-watt threshold on paper could have real-world ceilings 20 to 30 watts apart.
Same test. Different engine.
Other coaching sources report gaps as wide as 9% rather than the tidy 3% shown in that lab study, depending on the test protocol and the athlete tested. Either way, the lesson holds: don’t treat CP and FTP as interchangeable numbers. Pick one, and know why.
Why CP Runs Higher: The W’ Tank Explanation
Here’s the metaphor that makes this click. Think of your legs as running on two tanks. One tank is your aerobic engine, the power you can hold more or less indefinitely. That’s roughly what FTP and CP both try to measure.
The second tank is smaller and doesn’t refill mid-ride. Exercise scientists call it W’ (W-prime), a fixed reserve of hard, above-threshold work, usually somewhere around 15 to 25 kilojoules in a trained cyclist. Once it’s empty, you slow down no matter how badly you want to hold the pace. There’s no coasting your way to more of it.
The Critical Power model measures both tanks: the steady aerobic engine (CP) and the reserve tank (W’). A standard FTP test doesn’t separate the two. That’s part of why CP often reads a bit higher. It’s built to isolate the sustainable number, while a single 20-minute FTP test still has some of that reserve tank baked into the result.
The Hidden Trap: How Your 20-Minute Test Already Oversells You
This is the layer almost nobody talks about, and it happens before you ever compare CP to FTP.
A 20-minute all-out effort isn’t purely aerobic. It draws on that reserve tank too. For riders with a sprinter-type profile, meaning a big W’ and a lower aerobic ceiling, the anaerobic contribution to a 20-minute test can run large. In one analysis, roughly 24 watts of a sprinter’s 20-minute average came from anaerobic reserve rather than sustainable aerobic power. That drags the real ratio down closer to 90.8% of 20-minute power, not the standard 95% multiplier most coaches use.
Translation: if you’re a punchy, anaerobic rider, your 20-minute test already overstates your true threshold before anyone even mentions Critical Power.
It gets messier from there. Comparisons of different Critical Power calculation models, the 2-parameter version against 3-parameter versions, show that whichever model reports the highest CP tends to report the lowest W’, and the reverse holds too. Even “the CP number” isn’t one fixed answer. It shifts depending on which app or spreadsheet crunched it.
Two layers of inflation, stacked.
Ironman and 70.3 Bike Targets: The IF Numbers That Protect Your Run
Intensity Factor (IF) is your race power divided by FTP. Ride at 0.70 IF for four hours and you’re holding 70% of your one-hour max the whole time.
| Race Distance | Recommended IF | Who It’s For |
|---|---|---|
| Full Ironman, first-timer | 0.65-0.70 | New to the distance, conservative pacing |
| Full Ironman, experienced | 0.68-0.76 | Dialed-in racers with good heat tolerance |
| Ironman 70.3 | 0.76-0.84 | Shorter run afterward, more room to push |
For a full Ironman, most coaching guidance lands between 0.68 and 0.76 IF, with first-timers and conservative pacers closer to 0.65-0.70. A 70.3 bike leg allows more, typically 0.76-0.84 IF, because there’s a shorter, easier run waiting on the other side.
Notice these ranges sit comfortably below where a CP-derived number would put you. That gap is intentional. It’s your margin for the marathon.
Why Outdoor Racing Breaks the Clean Power Model
Critical Power models are built from smooth, steady lab tests. Real Ironman courses aren’t smooth. Wind gusts, rollers, and aid-station surges push your power around all day.
We measure that wobble with Variability Index (VI), the ratio of your Normalized Power to your plain average power. A VI of 1.00 means your power held dead flat. Higher numbers mean it bounced around more.
| Course Type | Target VI | What It Means |
|---|---|---|
| Flat Ironman or 70.3 course | 1.00-1.03 | Power barely swings; CP and FTP targets both hold up |
| Rolling or hilly course | 1.04-1.07 | Power swings more; lean on FTP, the safer number |
| Criterium or road race | 1.20-1.35 | Power swings wildly; a steady-state model doesn’t apply |
The higher your course’s expected VI, the less you should trust a razor-precise CP number for pacing. A steady-state model assumes steady output. Hills and gusts don’t care what your power meter says your threshold is.
The Run Blow-Up: What Overcooking the Bike Actually Costs You
Every watt you ride above what your body can truly sustain burns into your reserve tank and your glycogen stores. Neither comes back mid-race.
Coaches use a simple field test for this: if your first-hour average power beats your second-hour average by more than 5%, you went out too hard. That decay pattern is the fingerprint of a CP-inflated target dragging a rider past their real threshold.
The bill comes due on the run. Carbohydrate tolerance drops sharply as a race goes on. By roughly kilometer 21 of the Ironman marathon, most athletes can only stomach about half the carb intake rate they managed in the first hour of the bike. Burn your fuel early and there’s no refund window later.
Take a rider we’ll call Dave, 42, training for his second full Ironman. Dave has a sprinter’s power profile: strong short efforts, a big W’ tank, less diesel underneath. In his first Ironman, he pulled his bike number from a CP model and rode 265 watts, an IF that felt “right” on paper. He split a strong 5:10 bike. Then his run fell apart into a walk-jog shuffle and a 4:40 marathon.
For his second race, he rebuilt his target off FTP instead, riding a flat 230 watts (IF 0.70). His bike split slipped 10 minutes slower, to 5:20. His run came home in 3:52.
Same fitness. Different number to pace by. Thirty-eight minutes faster overall.
Fast bike, slow run isn’t bad luck.
Decision Framework: Which Number to Use for Your Profile and Distance
| Power Profile | Race Distance | Pace By | IF Target | Adjustment Note |
|---|---|---|---|---|
| Sprinter / high W’ | Ironman 70.3 | FTP | 0.76-0.80 | Subtract 0.02 if your FTP came from a CP20-style test |
| Sprinter / high W’ | Full Ironman | FTP | 0.65-0.70 | Use the low end; CP is least reliable for your profile |
| Balanced / all-rounder | Ironman 70.3 | FTP or CP | 0.78-0.84 | CP and FTP usually sit within 5W; either works |
| Balanced / all-rounder | Full Ironman | FTP | 0.68-0.74 | Stay mid-range |
| Diesel / TT-type | Ironman 70.3 | CP or FTP | 0.80-0.84 | Low W’ means CP tracks FTP closely |
| Diesel / TT-type | Full Ironman | FTP | 0.70-0.76 | Highest sustainable end of the range |
If you’re a sprinter-type rider with a big W’, trust FTP over CP for both distances, and lean toward the conservative end of the range, especially for a full Ironman. Your 20-minute test is the one most likely to be inflated in the first place. If you’re a diesel or time-trial type rider with a small W’, CP and FTP usually sit close together, though FTP still gives a safer margin on race day.
AthleteOS builds your full power-duration curve from every ride you import, then calculates both your FTP and a modeled Critical Power, with W’, from the same data. For long-course racing, it defaults your target to the FTP-based IF range rather than the CP number, and adjusts further for riders whose curve shows a high W’ contribution, exactly the profile most likely to get burned by an inflated 20-minute test. AthleteOS’s session analysis also checks your first-hour-versus-second-hour power split after every ride and flags it once the gap crosses 5%, so you catch an overcooked pacing plan before it costs you a marathon.
Your drift ratio tells you after the fact whether you paced it right. Getting your FTP test right in the first place matters just as much as which threshold model you race by, and watching your fitness and fatigue scores in the weeks before race day tells you which end of the IF range you can actually afford. Multi-sport athletes juggling TSS across three disciplines run into the same single-number trap: don’t let one metric from one sport set your pacing for the whole day. If you want AthleteOS to run these numbers for your next race, start a free account and upload your recent rides.