Sub-2 Marathon: How Sawe Ran 1:59:30

The barrier that had an asterisk

You may remember 2019. Eliud Kipchoge ran 1:59:40 in Vienna and the world cheered. But that run could never count as a world record, and everyone involved knew it. It used 41 pacemakers rotating in groups of seven, a car projecting a laser line on the road, drinks handed over by bicycle, and no open competition — and for exactly those reasons World Athletics could not ratify it (INEOS 1:59 Challenge; CBS Sports, 2019). It was a beautifully engineered demonstration that the human body could do it. It was not a race.

So the question stayed open for seven more years: could someone do it for real — in an open race, with a normal field, under the sport's ordinary rules?

Kelvin Kiptum's 2:00:35 in Chicago (2023) came agonisingly close and then stood untouched. Between 2023 and 2025 the world's best kept landing in the 2:00–2:02 band. The barrier looked like a wall.

Then London 2026 happened, and three men went through it in the same race.

The race: a masterpiece of patience

Here is what makes the run so strange. Sawe did not blast the first half and hang on. He did the opposite. The official 5 km splits (LetsRun; London Marathon Events):

Read the last two rows again. His fastest 5 km of the day came between 35 and 40 km — the part of a marathon where most runners are slowing down. First half: 60:29. Second half: 59:01. That is an 88-second negative split at world-record pace.

The final 2.195 km took 5:51, about 4:17 per mile (LetsRun).

And here's the detail that says the most: the pacemakers left just after 25 km, and Sawe still ran the second half faster than the first. He was on his own for the quickest 17 kilometres of the race.

He was not alone in the effort. Kejelcha sat on his shoulder until about 40 km before Sawe made his decisive move with one mile to go. Jacob Kiplimo was dropped near 30 km and still finished in 2:00:28 — also inside the old world record (Olympics.com; CITIUS Mag). Three men, one race, one wall.

Who are these two?

Sabastian Sawe (Kenya, born 16 March 1995) is almost absurdly new to this. London 2026 was his fourth marathon. He has won all four: Valencia 2024 in 2:02:05 on debut, London 2025 in 2:02:27, Berlin 2025 in 2:02:16, then this. Before that he was a half-marathon specialist with a 58:05 personal best and a world road running title (athlete profile). He trains at roughly 1,900–2,000 m altitude in Kapsabet, western Kenya, under Italian coach Claudio Berardelli. Berardelli has put real numbers on the build: over the final six weeks Sawe averaged 200 km a week and peaked at 241 km — nine or ten straight training days before a rest day (Daily Nation; Irish Times).

That build was not smooth. Berardelli has described a foot injury in autumn 2025 and a back injury in December that left him unsure Sawe could even race London, with full training resuming only in February — and said the disruption forced them to take risks in training to arrive in shape (LetsRun preview; RunnersConnect).

One more thing sets Sawe apart. Since 2025 he has volunteered for enhanced, independent testing by the Athletics Integrity Unit — roughly 25 tests a year, with his sponsor contributing $50,000 annually to fund it, and neither he nor his team told when or how the tests will come (LetsRun; Pulse Sports Kenya).

Yomif Kejelcha (Ethiopia, born 1 August 1997) came from the track: a former world indoor mile record holder, 26:31 for 10,000 m, twice a world championship silver medallist at that distance. London was his marathon debut — 1:59:41, the fastest first marathon in history. His fuelling was engineered station by station by a Spanish sports-nutrition firm, planned around roughly 60 ml of fluid at most stops (CITIUS Mag).

He did not get all of it. He skipped a bottle early and took nothing at 40 km — and by his own account, at 41 km his body "got stuck" and his legs "stopped". His nutrition team's calculation is that the missed carbohydrate may have cost him the final kilometre (CITIUS Mag). Hold that thought; it becomes important later.

Now the physics: what does 1:59:30 actually demand?

Everything below is arithmetic you can follow, and check, yourself. The only inputs are the finish time and published laboratory data on runners at this level.

Step 1 — How fast is it, really?

42,195 metres ÷ 7,170 seconds = 5.885 metres per second. That's 21.19 km/h, or 2:49.9 per kilometre (4:33.5 per mile).

Put differently: that is a lap of a 400 m track every 68 seconds, repeated 105 times without a break.

Step 2 — How much oxygen does that cost?

Running costs oxygen, and how much it costs you is your running economy — think of it as fuel consumption per kilometre. Elite runners burn less oxygen for the same speed than the rest of us.

In the best data we have on runners at this level — Andrew Jones and colleagues tested the world-class athletes screened for the Breaking2 project — the measured oxygen cost was 191 ± 19 ml of O₂ per kilogram of body weight per kilometre while running outdoors at roughly 21.1 km/h, almost exactly Sawe's race pace (Jones et al., 2021).

Apply that to Sawe's speed:

190 ml/kg/km × (21.19 ÷ 60) km per minute = 67.1 ml/kg/min

Modern carbon-plated shoes lower that cost by roughly 4% (Hoogkamer et al., 2018). If his effective economy was 180–185 ml/kg/km instead, the demand drops to roughly 63–66. Elite Ethiopian men have in fact been measured at 178–181 ml/kg/km at 16–18 km/h (BMC Research Notes, 2025) — so economy in that band is not a fantasy.

Central estimate: about 65–68 ml of oxygen per kilogram per minute, held for two hours. For scale: that sustained rate is roughly what 30 elite Ethiopian men produced as their maximum when tested at altitude — 66.2 ± 5.9 (BMC Research Notes, 2025).

Step 3 — Convert that into power and calories

One millilitre of oxygen releases about 20.9 joules of energy. So:

190 ml/kg/km × 20.9 J = 3,971 J/kg/km = 3.97 joules per kilogram per metre

That sits right on top of the classic measured value for trained runners — di Prampero's group measured 0.179 ± 0.017 ml O₂ per kg per metre above rest, about 3.86 J/kg/m, and found it did not change with speed (di Prampero et al., 1986). Multiply energy cost by speed to get metabolic power:

3.97 J/kg/m × 5.885 m/s = 23.4 watts per kilogram

With super-shoe economy, about 22.3. Call it 22–23.5 W/kg, sustained for two hours.

Total energy needs a body mass, and here we hit a hard limit: World Athletics publishes no height or weight for either athlete, so any figure is an assumption. Assuming a race weight near 55 kg:

23.4 W/kg × 55 kg × 7,170 s ≈ 9.22 million joules ≈ 2,200 kcal

Depending on his true race weight (52–58 kg), somewhere between 2,050 and 2,300 kcal — a full day's food, burned in two hours.

Step 4 — What share of his ceiling was that?

Your VO₂max is the maximum rate you can consume oxygen — the size of your engine. The share of it you can hold for two hours is called fractional utilisation.

If his race demand was ~67 ml/kg/min:

So: roughly 86–93% of his ceiling, for two hours.

That estimate has a real-world anchor. When Jones and colleagues put world-class runners on the road at this exact pace, the seven who could reach a steady state were working at 94 ± 3% of their peak oxygen uptake, needing about 4.0 litres of oxygen per minute (Jones et al., 2021). Sawe ran faster than that test pace, for the full distance. Either his engine is bigger than theirs, or his efficiency is better, or — most likely — both, by a little.

Step 5 — Work backwards to the engine

Now flip the equation. VO₂max = race demand ÷ fractional utilisation. With demand at 65–68 and utilisation at 86–92%:

VO₂max ≈ 70–80 ml/kg/min, most likely 74–77.

And here is the genuinely surprising conclusion: that is not a superhuman number. Plenty of elite distance runners test in that range, and exceptional amateurs occasionally reach the bottom of it. The Breaking2 group averaged 71.0 ± 5.7 ml/kg/min, with individuals spanning roughly 64–84 (Jones et al., 2021). Elite Kenyan runners measure around 63–70 at altitude and 73–85 at sea level (Larsen, 2003; Larsen & Sheel, 2015), and 30 elite Ethiopian men averaged 66.2 ± 5.9 when tested at altitude (BMC Research Notes, 2025).

A sub-2 marathon was never won by having the biggest engine. It was won by having a very good engine, running it at an extraordinary percentage of its limit, and wasting almost none of the fuel.

Step 6 — He ran the whole race at his red line

In that same Breaking2 group, the lactate threshold — where lactate first starts climbing — sat at 18.9 ± 0.4 km/h (83 ± 5% of VO₂peak), and the lactate turn point — the intensity above which fatigue stacks up fast, roughly the quickest speed you can hold in a steady state — at 20.2 ± 0.6 km/h (92 ± 3% of VO₂peak) (Jones et al., 2021).

Sawe ran 21.19 km/h for 42 kilometres.

He spent two hours at, or slightly above, the speed where the physiology says things should be falling apart — and then he accelerated. That single fact is the most extreme part of this performance, and no current model explains it well; the newest attempt to frame a record-eligible sub-2 explicitly folds in technology and cognitive factors alongside the classic three (2026 integrative framework).

The fuel problem nobody sees on TV

Take the ~2,200 kcal from Step 3 — about 1,100 kcal an hour. At an intensity this close to maximum, almost all of that comes from carbohydrate rather than fat. Carbohydrate yields roughly 4 kcal per gram, so the burn rate works out at 200–250 grams per hour, or 400–500 g across the race.

Here's the catch: the human gut can only absorb so much. Standard sports-nutrition guidance tops out near 90 g per hour, and only with glucose–fructose blends that use two separate intestinal transporters — single-sugar drinks cap out around 60 g/hr (Jeukendrup, 2014). Elite practice has since pushed that to roughly 120 g/hr with months of deliberate gut training.

So the sums don't balance. Burn 200–250 g an hour, absorb at best 120, and every hour you run a deficit that has to be covered by glycogen — the carbohydrate already stored in your muscles and liver before the gun. That store is finite, which is the entire reason carbo-loading exists. Run it down and you don't slow gracefully; you stop.

This is why Sawe averaged around 115 g of carbohydrate per hour — near the physiological ceiling, using hydrogels plus a bicarbonate buffer, on a plan refined over roughly twelve months of gut training that involved six dedicated trips to Kenya by his nutrition team (Marathon Handbook; NutraIngredients). And it's why Kejelcha's two missed bottles — a mere 19 g — showed up as an energy crisis at 41 km. Nineteen grams. In a race decided by eleven seconds.

The stack of small edges

No single thing broke two hours. Roughly a dozen small advantages arrived in the same place on the same morning.

Add them up — footwear worth perhaps 1.5–4%, ideal weather worth 1–2% over a warm day, course and pacing worth around 1%, plus modern fuelling and low body mass — and you can account for the 1–2 minute step from the 2:00–2:02 plateau to 1:59:30. The gap had already shrunk to seconds by 2025. London simply delivered every variable at once.

The prediction that came true 35 years early

In 1991, physiologist Michael Joyner built a simple model: marathon speed comes from three things multiplied together — engine size (VO₂max), the share of it you can hold (fractional utilisation), and efficiency (running economy). He plugged in the best plausible values for each and got a theoretical human limit of 1:57:58 — from "a hypothetical subject with a VO₂max of 84 ml/kg/min, a lactate threshold of 85% of VO₂max, and exceptional running economy" (Joyner, 1991).

Twenty years later, Joyner and colleagues went further and described who would do it first: someone with "outstanding running economy and small body size along with exposure to high altitude and significant physical activity early in life" — most likely an East African (Joyner, Ruiz & Lucia, 2011).

A Kenyan and an Ethiopian, both small, both trained at altitude, finishing eleven seconds apart. The prediction was almost embarrassingly accurate.

Note also what the model didn't require: a record-shattering VO₂max. Their edge is efficiency — including, research suggests, the slender lower legs that cost less energy to swing with every stride (Larsen & Sheel, 2015). And the measured oxygen cost for elite Ethiopian men lands at 162–181 ml/kg/km across 12–18 km/h (BMC Research Notes, 2025), comfortably below typical trained values. That study's own conclusion is the same one this race demonstrated: exceptional economy can compensate for a merely very good VO₂max.

What this means if you're not running 2:50 per kilometre

Strip out the elite context, and Sawe's race is a checklist of things that work at every speed:

1. Even or negative pacing. The fastest marathon in history was run 88 seconds quicker in its second half than its first — not blasted from the gun and defended.
2. Economy matters as much as engine. In the sub-2 cohort, no single variable predicted performance; VO₂max, the fraction of it sustained, and running economy only made sense in combination (Jones et al., 2021). The training that moves economy is covered, with its evidence, in Running Economy.
3. Fuel is a trainable skill. Your gut adapts to carbohydrate the way your legs adapt to mileage, and the ceiling depends on using mixed sugars (Jeukendrup, 2014). Practise it.
4. Conditions are worth minutes. Every 5°C of extra heat costs roughly 1% (Ely et al., 2007). Choose your goal race the way Sawe chose London.
5. Shoes are real but small. About 4% metabolically, ~2% in finish time (Kipp et al., 2019) — they tilt a result; they don't create one.

How much further can it go?

Joyner's theoretical 1:57:58 has now been approached, not reached — and nobody knows whether that figure is the true ceiling or just the best guess available in 1991. It is worth remembering how badly the forecasts have aged: the most careful statistical model of record progression placed a legal sub-2 in the early 2030s (Angus, 2019). It happened years ahead of schedule.

The nearer question is simpler. Sawe is 31, four marathons into his career, and ran his fastest 5 km segment of a world record in the 36th kilometre after a winter of injuries. Kejelcha ran 1:59:41 on his first attempt, missed fuel late, and still finished eleven seconds off the record.

Two hours was a wall for a very long time. It turns out it was a door.

References

Peer-reviewed research

1. Joyner MJ (1991). Modeling: optimal marathon performance on the basis of physiological factors. J Appl Physiol. 70(2):683–687.
2. Joyner MJ, Ruiz JR, Lucia A (2011). The two-hour marathon: who and when? J Appl Physiol. 110(1):275–277.
3. Jones AM et al. (2021). Physiological demands of running at 2-hour marathon race pace. J Appl Physiol. 130(2):369–379.
4. di Prampero PE et al. (1986). The energetics of endurance running. Eur J Appl Physiol. 55(3):259–266.
5. Larsen HB (2003). Kenyan dominance in distance running. Comp Biochem Physiol A. 136(1):161–170.
6. Larsen HB, Sheel AW (2015). The Kenyan runners. Scand J Med Sci Sports. 25(S4):110–118.
7. Hoogkamer W et al. (2018). A comparison of the energetic cost of running in marathon racing shoes. Sports Med. 48(4):1009–1019.
8. Kipp S, Kram R, Hoogkamer W (2019). Extrapolating metabolic savings in running: implications for performance predictions. Front Physiol. 10:79.
9. Ely MR et al. (2007). Impact of weather on marathon-running performance. Med Sci Sports Exerc. 39(3):487–493.
10. Jeukendrup AE (2014). A step towards personalized sports nutrition: carbohydrate intake during exercise. Sports Med. 44(S1):25–33.
11. Angus SD (2019). A statistical timetable for the sub-2-hour marathon. Med Sci Sports Exerc. 51(7):1460–1466.
12. Toward a record-eligible sub-2-hour marathon (2026). An updated integrative framework of physiological, technological and cognitive determinants.
13. Association between running economy and VO₂max in high-level Ethiopian distance runners (2025). BMC Res Notes. 18:s13104-025-07397-8. — 30 elite men, VO₂max 66.2 ± 5.9 ml/kg/min at altitude; economy 162–181 ml O₂/kg/km.

Race, rules and reporting

14. World Athletics (23 July 2026). Ratified: world records for Sawe, Ehammer, Charlton, McRae, Yan, Duplantis and Assefa. Press release.
15. LetsRun (26 April 2026). 1:59:30 — Sabastian Sawe shatters the 2-hour barrier at the 2026 London Marathon. (Splits, weather, race dynamics.)
16. LetsRun (May 2026). Did Sabastian Sawe really run a 4:12 mile? No, he did not.
17. London Marathon Events (2026). Sawe smashes two-hour barrier to make sporting history.
18. Olympics.com (2026) and CITIUS Mag. Race recaps.
19. AIMS / World Athletics. World record criteria for road races (elevation drop ≤1:1000; start–finish separation ≤50% of distance).
20. Ineos 1:59 Challenge and CBS Sports (2019). Why Kipchoge's 1:59:40 was not ratifiable.
21. Road Trail Run (2026) and Forbes (2026). Shoe specifications and manufacturer claims.
22. Marathon Handbook (2026) and NutraIngredients (27 April 2026). Sawe's ~115 g/hr fuelling plan.
23. CITIUS Mag (2026). Sawe and Kejelcha's fueling strategy to break two hours. (Kejelcha's station-by-station plan and missed bottles.)
24. Daily Nation (2026) and Irish Times (27 April 2026). Berardelli on the training build: 200 km/week average, 241 km peak.
25. LetsRun (April 2026) and RunnersConnect. Injury history and return to full training.
26. LetsRun (2026) and Pulse Sports Kenya (2026). Sawe's voluntary enhanced AIU testing programme.
27. Marathons.com (2026). Sawe career results and profile.

This article is for general education and isn't medical advice. The physiological values attributed to Sabastian Sawe and Yomif Kejelcha are modelled estimates derived from published research on comparable athletes, not laboratory measurements — no lab data has ever been published for either runner. If you're new to exercise, older, or managing a health condition, check with a clinician before starting or intensifying a training program.