Your pace hasn’t changed. Your heart rate has climbed 15 beats. You’re past mile 21 and your body is running above its own lactate threshold without you realizing it.
That’s the dual squeeze. And it explains most blown marathons.
A 2025 study published in the Scandinavian Journal of Medicine & Science in Sports measured what happens to the three pillars of endurance performance — VO2peak, threshold speed, and running economy — during a real long run. All three decline. Simultaneously. By the time well-trained male marathoners hit 120 minutes of threshold-pace running, their VO2peak had dropped 7.1%, their speed at lactate threshold had fallen 6.6%, and their running economy was 5.8% worse.
Fresh-state lab data doesn’t show any of that.
The VO2max Ceiling Drops While the Floor Rises
Here’s what the Zanini et al. (2025) study actually found in 14 well-trained male marathoners averaging a 2:46 finish time.
At the start of a run, these athletes were working at about 79% of their VO2peak. At 90 minutes, that fraction had climbed to 85.6%. By 120 minutes, it hit 90.6%. That means the effort was getting progressively harder relative to their ceiling.
But the ceiling was also dropping.
VO2peak fell 3.1% at 90 minutes and 7.1% at 120 minutes. Translation: they were working at a higher percentage of a smaller engine. Both effects hit at the same time, compounding each other.
Think of it like a fuel pump drawing from a tank that’s slowly draining. The pump has to spin faster just to deliver the same flow. That’s your heart rate climbing at constant pace. And while that’s happening, someone is also shrinking the tank itself.
Notice how both lines converge. That convergence is the physiological definition of a death march.
How Lactate Threshold Speed Decays Under Race Conditions
Lactate threshold speed is the fastest pace you can run before lactate starts accumulating faster than your body can clear it. It’s the most important number in marathon pacing. And it isn’t fixed.
In the Zanini 2025 data, threshold speed started at 14.0 km/h (4:17/km). At 90 minutes into a run, it had dropped to 13.5 km/h (4:26/km). By 120 minutes, it was 13.0 km/h (4:37/km). That’s a 20-second-per-km swing at a pace the athlete never changed.
The Hunter and Muniz-Pumares 2025 study found the same pattern in London Marathon runners. Threshold speed dropped 5.5% after 90 minutes of running, from 12.8 to 12.1 km/h. More importantly, the size of that drop correlated strongly with actual race performance (r = 0.68). Athletes who held their threshold under fatigue ran faster. Athletes who didn’t, fell apart.
The lab number your coach has from your fresh-state test isn’t the number you’ll race on. It’s the best-case version.
The 26 km Tipping Point
If you’ve run a marathon and “hit the wall,” there’s population-level data describing exactly when it happens.
In a study of 82,303 recreational marathon runners (Smyth, Maunder et al., Sports Medicine 2022), aerobic decoupling began on average at 25.2 km. The drift ratio — the gap between how hard the heart is working versus actual pace — averaged 16% across the full race. And 77% of runners experienced a significant speed drop starting at the 26th kilometer.
Runners with a drift ratio above 1.2 (meaning internal load increased 20% or more relative to pace) finished 21 minutes slower than runners who stayed under 1.1. They also started decoupling at 19.1 km, not 33.4 km. They hit the wall 14 km earlier.
That gap of 21 minutes doesn’t come from fitness differences measured in a lab. It comes from durability.
The watch isn’t lying. The drift ratio is the real signal.
For a deeper look at how decoupling works and how to read your own numbers, see aerobic decoupling explained. To understand the specific mechanics of why runners hit the wall at mile 20, the substrate and pacing story goes deeper.
The Substrate Shift Nobody Talks About
One mechanism behind threshold decay is fuel. As glycogen depletes, the body shifts toward fat oxidation. That sounds fine in theory. In practice, it costs you.
The Hunter 2025 study tracked fuel use. Carbohydrate oxidation fell from 2.9 g/min to 1.9 g/min across the 90-minute run. Fat oxidation roughly doubled, from 0.26 to 0.52 g/min. Fat is a slower, less efficient fuel per unit of oxygen consumed. So at the same pace, you’re now spending more oxygen to produce the same amount of energy.
Running economy deteriorated 4.2% at 90 minutes and 5.8% at 120 minutes in the Zanini 2025 data. A 6% economy decline means you’re burning roughly 6% more oxygen per stride. Combined with a 7.1% drop in how much oxygen you can take in, you’re in a two-way deficit. Same stride, shrinking budget, leaking tank.
This is why gel timing matters at mile 18, not mile 22. You can’t fully reverse the substrate shift, but you can slow it. Good fueling keeps carbohydrate oxidation higher for longer, which delays threshold decay. For more on the science of fueling long efforts, see how Zone 2 training builds your aerobic base.
One Runner’s Before and After
Take a runner I’ll call James. Forty-one, training for his second marathon, running about 55 miles per week. His lab threshold speed tested at 12.4 km/h. He paced his first marathon on that number. He hit 28 km in good shape. Then his heart rate climbed 14 beats in 6 minutes without pace change, and his last 14 km took 18 minutes longer than projected.
His drift ratio on that race was 1.24 — high decoupling territory.
Over the next 16 weeks, he added two things. First, fast-finish long runs: the final 30 minutes at goal marathon pace rather than easy pace. Second, two weekly strength sessions including heavy squats and plyometric bounding. In his next marathon, his drift ratio dropped to 1.07. He hit the same 28 km mark at the same pace, and the heart rate barely moved. He finished in 3:38 — 19 minutes faster.
His VO2max didn’t change. His durability did.
Training Durability: What Actually Works
Two 2025 studies point to specific solutions.
The first is the Zanini, Folland, Wu, and Blagrove RCT published in Medicine & Science in Sports & Exercise. Fourteen well-trained male runners added twice-weekly maximal strength and plyometric training for 10 weeks. Running economy durability at 90 minutes improved 2.1% in the intervention group versus a 0.6% deterioration in controls. Post-fatigue time-to-exhaustion improved 35% in the strength group, compared to an 8% decline in controls.
That’s not a marginal benefit. Thirty-five percent more capacity at the end of the race.
The second is a Frontiers in Physiology study from Meyler and colleagues (2023) showing both low-intensity and high-intensity training can delay the onset of physiological drift. Before training, drift started around 107 minutes. After 10 weeks, it started at 134 minutes. Both groups improved similarly, with high-intensity training postponing drift onset from 108 to 137 minutes, and low-intensity from 106 to 131 minutes.
Translation: durability is trainable. And you don’t need a new lab number to prove it’s working.
Practical sessions that build it:
- Fast-finish long runs — hold easy pace for 70–80% of the run, then shift to marathon goal pace for the final 20–30 minutes.
- Pre-fatigued threshold blocks — run your threshold intervals in the back half of a 90-minute run, not fresh.
- Twice-weekly strength — heavy compound lifts and plyometrics, matched to the Zanini 2025 protocol cadence.
The strength component in particular is often the missing piece for self-coached runners. Most threshold training builds a ceiling. It doesn’t reinforce the floor. See how to combine strength and endurance training for a practical concurrent-training framework that won’t compromise your run volume.
What This Means for Your Pacing
Using the Zanini 2025 data to do the pacing math is sobering.
A 2:47 marathoner with a fresh threshold speed of 14.0 km/h should expect that number to read 13.5 km/h at mile 16 and 13.0 km/h at mile 21. Any pacing strategy targeting the original 14.0 km/h across the full race forces the athlete above their decayed threshold for the final 10-plus miles.
The fix is to build in a buffer. Runners with drift ratios above 10% should target a first-half pace 2–3% slower than their fresh threshold. That 2–3% sacrifice in the first half prevents the 15–20% collapse in the second half.
Your fresh VO2max gets you to the start line. Your durability determines where you finish.
AthleteOS calculates your drift ratio after every qualifying long run and surfaces a Durability Score showing how your pace-to-heart-rate coupling degrades under fatigue. When the score trends down across back-to-back long runs, your AI coach flags it and prescribes the right corrective session before your goal race. You can set up drift ratio tracking here with data from any Garmin or Strava-connected device.
Steady is faster than fast-then-slow. The numbers now prove exactly why.