If your heart rate climbs 12% over the second half of a long run while your pace holds steady, your aerobic base isn’t ready for race day. That gap, the drift between your internal load (heart rate) and your external output (pace or power), is what aerobic decoupling measures. It’s one of the sharpest predictors of endurance performance we have.
Smyth and colleagues (2022) tracked 82,303 recreational marathoners. Runners with low drift finished 21 minutes faster (217 vs 238 minutes). They held their pace until kilometer 33.4 before fading, compared to kilometer 19.1 for the high-drift group. That’s the difference between a strong finish and a long, painful walk to the line.
Aerobic Decoupling: The Formula and What It Actually Measures
Aerobic decoupling compares how efficiently you move in the first half of a workout to the second half. Efficiency is expressed as the Efficiency Factor (EF):
For cycling:
EF_cycling = (Normalized Power) / (Average HR)
For running:
EF_running = (Normalized Graded Pace) / (Average HR)
The drift ratio is then:
Drift Ratio = (EF_H1 - EF_H2) / EF_H1 * 100
In plain English: if you’re getting fewer watts per heartbeat (or less pace per heartbeat) in the second half, your engine is working harder to deliver the same output. Same throttle, less power.
Coach Joe Friel built this framework. It only works on long, steady, sub-threshold efforts. Intervals, hilly group rides, and races don’t qualify. The signal needs a flat baseline to read clearly.
What the Numbers Mean at the Race-Day Level
The thresholds aren’t arbitrary. They’re grounded in the same data Smyth et al. used to separate runners by finish-time group, and in Friel’s decade of coached athlete experience.
| Metric | Strong Base (< 5%) | Developing (5–10%) | Underdeveloped (> 10%) | Ironman Target |
|---|---|---|---|---|
| Cycling Pw:Hr (2–4 hr AeT effort) | Race-ready | More base needed | Intensity too high or base insufficient | < 5% over 4 hr |
| Running Pa:Hr (1–2 hr AeT effort) | Race-ready | More base needed | Intensity too high or base insufficient | < 5% over 2 hr |
| Marathon decoupling onset (km) | ≥ 30 km | 20–29 km | < 20 km | — |
| Marathon finish time (Smyth 2022) | 217 min avg | 225 min avg | 238 min avg | — |
The Ironman-specific benchmarks are more demanding: you need to sustain < 5% Pw:Hr for a full 4-hour ride at aerobic threshold, then repeat the test with a < 5% Pa:Hr 2-hour run. Anything less and your base isn’t ready for 226 km of racing.
Understanding this metric connects directly to building the aerobic foundation that drives it. Zone 2 training is the primary lever. It’s where the mitochondrial adaptations that reduce drift actually happen.
Why Drift Happens: A Fuel Pump Drawing from a Draining Tank
Two things stress your heart on a long, hot effort. They look small alone. Together, they hit much harder.
When you run long in heat, you sweat off plasma volume and your core temperature climbs. Your heart beats faster to keep blood moving, but each beat pumps less. The stroke volume (blood ejected per beat) drops.
Now the painful part. González-Alonso and colleagues (1997) tested 15 endurance athletes one stressor at a time, then both at once. Heat alone (1°C core rise) dropped stroke volume by 11 ml/beat. Dehydration alone (4% body weight loss) also dropped it by 11 ml/beat. You’d expect roughly 22 ml/beat combined. The actual answer was 26 ml/beat, with a 13% fall in cardiac output. Not additive. Multiplicative.
Think of your heart as a fuel pump feeding an engine. Dehydration shrinks the fuel tank. Heat makes the engine spray fluid back at the radiator to cool itself. Either one alone, you can manage. Together, the pump is drawing from a draining tank while losing fluid out the back. The engine doesn’t stall. It just works a lot harder to deliver the same power.
That’s your heart rate climbing while your pace stays flat.
| Stressor | Stroke Volume Reduction | Cardiac Output Impact |
|---|---|---|
| Hyperthermia alone (1°C rise) | − 11 ± 3 ml/beat | Moderate |
| Dehydration alone (4% body weight) | − 11 ± 3 ml/beat | Moderate |
| Combined hyperthermia + dehydration | − 26 ± 3 ml/beat | − 13% (− 2.8 l/min) |
Source: González-Alonso et al., Journal of Applied Physiology, 1997
Montain and Coyle (1992) found this relationship is almost linear. Across 1.1% to 4.2% body weight loss in 8 cyclists, the correlation between dehydration and both HR rise and stroke volume drop was r = 0.99. Lose 1% body weight, expect a clear cardiovascular cost. Lose 4%, expect four times that cost.
So a hot session will always read higher drift than the same workout in cool weather. Don’t compare them. Test on similar days, or read the trend across weeks instead of session to session.
A Real Training Block: Maria’s 8-Week Drift Ratio
Maria is 38, training for a sub-3:30 marathon. She’s been logging 55 km/week for three months, mostly at “medium effort,” which turns out to be Zone 3. Her first aerobic decoupling test on a 90-minute easy run comes back at 9.2%. HR starts at 142 bpm and finishes at 156 bpm. She’s not going fast, but she’s leaking efficiency by kilometer 7.
She switches to eight weeks of Zone 2 base work, capping every run at 75% of max HR, with one long run a week and everything else kept genuinely easy. She tracks her weekly load with her fitness, fatigue, and form scores so she doesn’t ramp too fast.
By week 8, her same 90-minute test comes back at 4.5%. HR starts at 141 bpm and finishes at 148 bpm. Same effort. Same course. Same pace. Her cardiovascular system is now far more economical at that intensity. Better still, her decoupling onset shifted: drift didn’t appear until the final 20 minutes versus starting at kilometer 7.
That’s the adaptation. It happened not by training harder, but by training smarter.
When Aerobic Decoupling Runs Backward: The Elite Inversion
Here’s a finding most coaches miss. Takayama and Aoyagi (2025) studied elite marathoners finishing in 2:03 to 2:08 and found something that flips the standard model.
For these athletes, heart rate doesn’t drift upward. It can’t. They’re already running at near-maximal HR from kilometer 1, with no headroom left to climb. So when they fade, what shifts? Their pace. Pace slows, heart rate holds. The formula still picks up the efficiency loss, but the mechanism runs in reverse.
For recreational and national-level runners running below 90% of max HR early in a race, the standard model fits cleanly. If you’re coaching elites, watch pace, not HR.
Don’t compare your numbers to an elite’s anyway. It’s like comparing a sedan’s fuel economy at 60 mph to a Formula 1 car’s at 190. Different engines, different rules.
Running vs. Cycling: Why 6% Means Different Things
Pa:Hr (running) and Pw:Hr (cycling) share the same formula but don’t share the same interpretation scale.
Running involves eccentric muscle loading that cycling doesn’t. Every footstrike creates micro-damage that accumulates over a long run and independently degrades running economy. Your pace per heartbeat can fall in the second half not because your heart is struggling, but because your muscles are mechanically less efficient.
That inflates Pa:Hr readings compared to Pw:Hr readings for the same cardiovascular fitness. Friel’s test durations reflect this: 2–4 hours for Pw:Hr, 1–2 hours for Pa:Hr. Anchoring the test at the right absolute intensity matters too, which is where FTP comes in for cyclists.
For Ironman athletes trying to validate race-readiness, both tests matter, but they’re measuring overlapping, not identical, things. A 6% Pa:Hr on a 2-hour run is less alarming than a 6% Pw:Hr on a 2-hour ride, because some of that running signal is muscle damage, not cardiovascular drift.
The Trend Matters More Than Any Single Reading
One aerobic decoupling reading is noise. It depends on temperature, sleep, hydration, terrain variability, and the quality of your GPS data. What can’t be explained away is a consistent trend across 8–12 weeks.
AthleteOS calculates your drift ratio after every workout you import from Garmin or Strava. It splits the session at the midpoint, runs the math, and plots the trend across your training block. So a single noisy reading on a hot or tired day doesn’t send you into a spiral. The trend tells the story.
The Smyth 2022 study found that adding both the magnitude and the onset point of decoupling to marathon prediction models cut prediction error from 6.45% to 5.16%, a 20% improvement. The onset point matters independently. As base fitness builds, drift onset shifts later in the workout, often visible before the magnitude percentage drops at all.
If you want to start tracking this systematically, sign up for AthleteOS and connect Garmin or Strava. The decoupling trend appears automatically for every qualifying session. Cross-reference it against your training stress scores and you’ll see whether high-drift sessions correlate with high-load weeks. They almost always do.
For most athletes, three things are true:
- Drift ratio consistently above 10%? You’re training above aerobic threshold on your long sessions. Slow down.
- Stuck between 5% and 10%? You have a developing base. More volume, consistent zone, 8–12 weeks.
- Below 5% at target-distance? Your aerobic engine is ready. That’s when adding intensity starts making sense.
The number is simple. The discipline to act on it is the harder part.