Your marathon pace calculator gave you 4:45 per km. You held it through 28 km, then watched the wheels come off. The problem wasn’t your fitness. It was that your threshold pace at km 35 isn’t the same number you tested fresh on a Tuesday morning.
Lactate threshold speed drops 6.6% after just two hours of running, even in well-trained athletes. Standard race calculators assume it stays flat for three-plus hours. It doesn’t.
What Threshold Pace Durability Actually Means
Your lactate threshold (LT) is the pace at which lactate starts building in your blood faster than your body can clear it. It’s the ceiling on how hard you can run for a long time.
But that ceiling isn’t fixed. It shifts downward the longer you run.
Think of your aerobic system as a fuel pump drawing from a slowly draining tank. Early in the race, the pump pulls easily and delivers full pressure. An hour in, the tank is lower. The pump has to work harder to deliver the same flow. That’s your heart rate climbing while pace holds flat, and it’s why your threshold starts sliding.
Threshold pace durability is how much your LT pace drops over time. A highly durable runner might lose only 1–3%. A typical recreational runner loses 5–10%. Most race calculators don’t know your number. They assume you’re the durable one.
The Research: How Much Does Threshold Pace Drop During a Marathon?
Three studies give us the clearest picture of how far LT speed falls with time on feet.
Zanini, Folland and Blagrove (2025, Scandinavian Journal of Medicine and Science in Sports) tested well-trained male marathon runners before and after controlled runs. Speed at the lactate threshold dropped 3.0% after 90 minutes (14.0 to 13.5 km/h) and 6.6% after 120 minutes (14.0 to 13.0 km/h). Running economy also worsened 5.8% by the two-hour mark.
Two hours in, even trained runners are running at a threshold nearly 7% lower than what they measured fresh.
Nuuttila et al. (2024, European Journal of Applied Physiology) found similar numbers in recreational runners. After 90 minutes at an easy pace, LT speed fell 5.8% in females and 5.3% in males — both statistically significant. These weren’t hard efforts. Just 90 minutes of comfortable running was enough to move the ceiling.
The cycling data confirms the pattern. Stevenson and colleagues found that threshold power at the first ventilatory threshold (VT1) dropped exactly 10.0% after two hours of moderate cycling — from 217 W to 196 W, with heart rate at that same threshold rising from 142 to 151 bpm. See the full cycling data at PubMed.
That 10% cycling figure is where the “decay rule” gets its name. It’s the floor for what a trained endurance athlete can expect to lose from their threshold over two hours of racing.
The 10% Decay Rule and Why Your Threshold Pace Calculator Is Wrong
VDOT is Jack Daniels’ system for setting training paces from race times. It’s useful. But it carries a hidden assumption.
Marathon pace in the VDOT system sits at roughly 80–84% of VO2max. That’s calibrated on runners whose threshold holds near-flat for three hours. That describes elite athletes. It doesn’t describe most recreational runners.
About 75% of recreational runners run slower than their VDOT-derived marathon prediction when projecting from a half-marathon time. The formula isn’t broken. The durability assumption behind it is just wrong for most people.
The practical correction: if your fresh-state threshold suggests a 4:00/km marathon pace, your effective threshold at km 35 is likely somewhere between 4:12 and 4:24/km. That 5–10% gap is the difference between a clean race and a death march.
77% of Sub-Elite Runners Are “Fallers” — Most Don’t Know Until Kilometer 26
Billat and colleagues studied 280 sub-elite marathoners with finish times between 2:30 and 3:40. 77% were “fallers” — runners who slowed by at least 10% from their first 4 km segment to their last. Only 23% held pace.
The fallers averaged 3:01:42. The non-fallers averaged 2:54:09. That’s a 7-minute gap between runners with similar fitness who handled pace decay differently.
What made it worse: heart rate began drifting upward at km 21, well before speed collapsed at km 26. The body was already sending the signal.
A separate analysis of 82,303 recreational marathon runners found the average internal-to-external workload ratio was 1.16, meaning heart rate inflated 16% relative to pace by the end of the race. Runners with a drift ratio above 1.2 finished 21 minutes slower on average and sustained only 82% of their critical speed in the final 5 km. Low-decoupling runners sustained 88%. (Smyth et al., Sports Medicine 2022)
Your drift ratio doesn’t lie, even when your ego does.
The Individual Durability Gap Is Enormous
Here’s what most coaches don’t mention: the variation between individuals is massive.
Andrew Jones, writing in the Journal of Physiology in 2024, describes durability as the “fourth dimension” of endurance performance — independent of VO2max, lactate threshold, and running economy. In competitive cyclists, critical power decline after a hard 1,000 kJ effort ranged from less than 1% to approximately 32% across individuals. Some athletes are nearly immune. Others lose a third of their threshold before the race ends.
That same spread exists in runners. Two athletes with identical VO2max values and identical fresh-state threshold paces can finish 20 minutes apart if one loses 2% of their LT speed and the other loses 10%.
Knowing your VO2max doesn’t predict your marathon time. A calculator that ignores durability will be wrong for most of the people who use it.
A Concrete Example: How Durability Shows Up in Real Training
Take a runner I’ll call James — 43, targeting his first sub-3:30 marathon, averaging 55 miles per week. His 30-minute threshold test put his LT pace at 4:45/km. His VDOT gave him a 4:58/km marathon pace, projecting to a 3:30 finish.
In training, his drift ratio on 90-minute long runs was 8.2%. That number meant his effective threshold at the 2-hour mark was probably closer to 5:14–5:20/km. His goal pace was faster than his late-race threshold.
After eight weeks of dedicated Zone 2 base building, his drift ratio on the same 90-minute effort dropped to 4.8%. He ran a 3:34 on race day — slower than his VDOT target, but a clean negative split with pace holding in the final 10 km. His next cycle, with a corrected goal pace, went to 3:28.
James didn’t need more fitness. He needed an honest durability number.
How to Measure Your Threshold Pace Durability Without a Lab
You don’t need blood lactate strips. You need a pace-to-heart-rate ratio over time.
The drift ratio (also called Pa:Hr, or the aerobic decoupling metric) compares your pace-to-heart-rate efficiency in the first half of a session to the second half. If efficiency drops more than 5% between halves, your aerobic system is fading faster than the effort warrants.
The benchmark, backed by coaching research and confirmed in the Smyth data: under 5% drift is good. Over 5% means either the pace is above your aerobic threshold or your durability needs work.
Run a 90-minute easy effort. Your watch can show you the heart-rate and pace data; any training log will let you calculate the ratio. A drift ratio under 5% at that duration means your LT is holding steady. Above 5% means the decay is happening before you’ve even reached race distance. Learn more about how to read cardiac drift as a training signal.
To build durability, the prescription isn’t complicated:
- More Zone 2 volume. Long, genuinely easy runs where the drift ratio stays low.
- Field-test your lactate threshold regularly, not just once a cycle.
- Set your training zones from a measured threshold, not from a formula.
- Run your long runs long enough to actually stress the durability window (90+ minutes minimum).
How AthleteOS Corrects Your Marathon Goal Pace
Standard tools give you one pace. AthleteOS gives you a corrected one.
After every threshold run and long run, AthleteOS calculates your drift ratio from the session data. Over time, it builds a picture of how your heart-rate-to-pace efficiency decays with duration: your personal LT durability profile.
If your drift ratio runs above 5% on 90-minute efforts, your marathon goal pace gets adjusted downward. The AI coach doesn’t wait until race day to tell you the target was wrong. It flags the problem in training, when there’s still time to build your aerobic base or recalibrate the goal.
Athletes with strong durability see their pace targets validated. Athletes with high drift see their targets corrected, often by 15 to 25 seconds per km. That correction is uncomfortable to see. It’s far less uncomfortable than blowing up at km 30.
Start tracking your drift ratio in AthleteOS before your next marathon build.
Steady is faster than fast-then-slow. The decay rule just tells you how much slower “steady” needs to be.
Further reading: