How Temperature Affects Marathon Pace: A Data-Driven Analysis

Of every weather variable a marathon runner will face on race day — wind, rain, humidity, altitude — temperature exerts the single largest influence on finishing time. The relationship between ambient temperature and endurance performance has been studied extensively, and the data is unambiguous: running a marathon in 25°C weather is a fundamentally different physiological challenge than running one in 10°C weather, even if your fitness is identical.

What makes temperature particularly treacherous is that its effect on pace is non-linear. A five-degree increase from 10°C to 15°C barely registers. The same five-degree jump from 25°C to 30°C can add ten or more minutes to a four-hour marathon. Cool conditions are ideal, but extreme cold introduces its own costs — muscle stiffness, respiratory cold stress, and impaired coordination. The performance curve is a shallow U-shape with a sharp upward tail on the warm side.

At Havik, we model this relationship for every race in our database. Our analysis engine applies temperature-adjusted pace predictions at the segment level — accounting for the fact that a race starting at 7°C in the morning may finish at 18°C by midday. This article breaks down the science, presents the data, and explains exactly how much time temperature will cost you (or save you) on race day.

The Science: The Ely et al. Temperature-Performance Curve

The foundational research on temperature and marathon performance comes from Ely et al. (2007), “Impact of Weather on Marathon-Running Performance,” published in Medicine & Science in Sports & Exercise. The study analyzed finishing times from 1.8 million marathon results across major U.S. marathons over a 36-year period (1970–2006) and correlated them with race-day weather conditions.

The key finding: the optimal ambient temperature for marathon performance is approximately 10–15°C (50–59°F). At this range, the human body can dissipate metabolic heat through convection and evaporation without significant cardiovascular strain. Below this range, performance degrades slightly due to increased muscle viscosity and respiratory cold stress. Above it, the degradation is steep and exponential.

Havik’s pace model implements this curve directly. We set the physiological optimum at 7.5°C — the midpoint of the 5–10°C band where the best marathon world records have been set — and calculate slowdown percentages that increase linearly in the cold and exponentially in the heat, consistent with the Ely curve.

Here is how the slowdown breaks down by temperature band:

• Below 5°C (41°F): +1–3% slowdown. Muscles stiffen, lung irritation from cold air increases breathing effort, and vasoconstriction reduces blood flow to working muscles. Manageable with proper layering, but not free.
• 5–10°C (41–50°F): 0–1% slowdown. Near-optimal. The body generates substantial metabolic heat during a marathon (~1,000 watts at race pace), and cool air efficiently removes it. World records happen here.
• 10–15°C (50–59°F): 0% (baseline). The textbook ideal range. Thermoregulatory load is minimal, sweat rate is moderate, and cardiac output stays fully available for propulsion.
• 15–20°C (59–68°F): +1–3% slowdown. Heat stress begins. Blood is diverted from working muscles to the skin for cooling. Heart rate creeps upward at the same pace. Most runners perceive this as “warm but fine” — until kilometer 30.
• 20–25°C (68–77°F): +3–6% slowdown. Significant heat stress. Sweat rate rises sharply, glycogen depletion accelerates, and cardiac drift becomes pronounced. This is where many runners hit the wall earlier than expected.
• 25–30°C (77–86°F): +6–10% slowdown. Severe conditions. Core temperature regulation becomes the primary limiter, not aerobic capacity. DNF rates climb sharply.
• Above 30°C (86°F): +10–15%+ slowdown. Dangerous territory. Medical tent visits spike. Completing the race becomes the goal, not a time goal.

The critical insight from the Ely data is that the curve is asymmetric. Cold slows you down gently and predictably. Heat slows you down exponentially — and the exponential bend starts around 20°C, well before most runners would call it “hot.”

The Data: Temperature Impact by Range

The table below translates the Ely curve into concrete numbers. We show the expected pace slowdown as a percentage and the projected finishing time for a runner whose baseline (optimal-weather) marathon is 4:00:00.

Notice the asymmetry. The cold penalty tops out around +3%. The heat penalty starts at +3% and climbs to +15% or more. For a 3:30 marathon runner, racing in 28°C instead of 12°C could mean finishing closer to 3:51 — a 21-minute difference from temperature alone, before accounting for humidity or wind.

Why Heat Hurts More Than Cold

The asymmetry in the temperature curve is not arbitrary. It reflects a fundamental constraint of human thermoregulation: the body is significantly better at generating heat than dissipating it.

1. Thermoregulatory Demand and Cardiac Drift

When ambient temperature rises above the thermoneutral zone during exercise, the body must divert blood flow from working muscles to the skin for convective and evaporative cooling. This is a zero-sum game: every liter of blood sent to the skin is a liter not delivering oxygen to your quads. The result is cardiac drift — heart rate rises progressively even at a constant pace, because stroke volume drops as blood pools in dilated skin vessels. At 25°C, a runner maintaining a steady 5:00/km pace might see heart rate climb from 155 bpm at kilometer 10 to 172 bpm at kilometer 35, without any change in speed.

2. The Dehydration Cascade

Sweat rate increases exponentially with temperature. At 10°C, a 70 kg runner might lose 0.6–0.8 liters per hour. At 25°C, that figure climbs to 1.2–1.8 liters per hour. A 2% loss in body mass from dehydration (just 1.4 kg for a 70 kg runner) is enough to measurably impair performance. At 4% loss, the cardiovascular system is under severe strain: blood viscosity increases, further reducing stroke volume and amplifying cardiac drift.

3. Accelerated Glycogen Depletion

Heat increases reliance on carbohydrate metabolism relative to fat oxidation. Research by Febbraio et al. (1994) demonstrated that muscle glycogen utilization is 15–25% higher at elevated muscle temperatures compared to normal conditions. The practical consequence: the wall — that dreaded point where glycogen stores run out and pace collapses — arrives earlier. In cool conditions, a well-fueled runner might hit glycogen depletion at kilometer 33–35. In 25°C+ heat, it can come as early as kilometer 28–30.

4. Central Governor Response

The brain monitors core temperature and will reduce motor unit recruitment before core temperature reaches dangerous levels (~39.5°C). This manifests as a feeling of “the legs won’t go” even though the muscles are not locally fatigued. It is a protective mechanism, and it is not something you can override with willpower. The hotter the conditions, the earlier this governor kicks in.

5. Why Cold Is Comparatively Benign

Cold has a limited impact on marathon performance for a simple reason: running generates enormous metabolic heat. A marathon runner produces roughly 800–1,200 watts of heat at race pace. Even at 0°C, this internal furnace keeps core temperature elevated within minutes of starting. The remaining cold-specific penalties — increased muscle viscosity, airways cooling, and slightly higher oxygen cost of breathing cold air — are small and can be partly offset by appropriate clothing. This is why the cold side of the curve is nearly flat compared to the heat side.

Humidity: The Hidden Multiplier

Temperature alone does not tell the full story. A 22°C race in Berlin (dry continental air) feels very different from a 22°C race in Miami (subtropical humidity). The missing variable is dew point — a far better predictor of running performance than relative humidity.

Why Dew Point, Not Relative Humidity?

Relative humidity is misleading because it is temperature-dependent. A relative humidity of 80% at 8°C is comfortable. The same 80% at 28°C is dangerous. Dew point, by contrast, is an absolute measure of atmospheric moisture. It directly predicts how effectively your sweat can evaporate — which is the body’s primary cooling mechanism during exercise.

Havik’s pace model uses a dew point threshold of 13°C (55°F). Below this threshold, humidity has negligible impact on performance. Above it, we apply a ~1.5% slowdown per 5.5°C increase in dew point. Here is the breakdown:

• Below 12°C dew point: Minimal impact. Sweat evaporates efficiently. This is the dew point range at most European marathons.
• 12–18°C dew point: Moderate impact (+1–3%). Evaporative cooling is impaired. Perceived effort rises. Typical of spring marathons in the eastern United States.
• Above 18°C dew point: Severe impact (+3%+). Evaporative cooling partially fails. The body cannot shed heat fast enough, and core temperature rises rapidly. Common at marathons in the Gulf states, Southeast Asia, and summer races worldwide.

WBGT: The Gold Standard for Heat Risk

The most comprehensive heat stress metric is WBGT (Wet Bulb Globe Temperature), which combines dry-bulb temperature, humidity (via wet-bulb temperature), and solar radiation (via globe temperature) into a single index. Havik calculates WBGT for every segment of every race analysis using the Stull (2011) wet-bulb approximation formula.

The American College of Sports Medicine (ACSM) recommends the following WBGT thresholds for endurance events:

• Below 18°C WBGT: Low risk. Normal racing conditions.
• 18–23°C WBGT: Moderate risk. Increased vigilance for heat illness.
• 23–28°C WBGT: High risk. Race organizers should consider schedule modifications.
• Above 28°C WBGT: Extreme risk. ACSM recommends cancellation of endurance events for non-elite participants.

When Havik detects WBGT above 23°C on any segment of a race, we flag it as a heat stress zone and adjust pace recommendations accordingly. This is the same threshold used by World Athletics for their heat mitigation policies.

Real Race Examples from Havik’s Database

Theory is useful, but runners choose specific races. Here is how temperature affects performance at six of the world’s most popular marathons, based on historical weather data from the past decade.

Berlin Marathon — The Weather Benchmark

Average race-day temperature: 14°C. Estimated pace impact: +0.8%. Berlin is not just fast because it is flat — it is fast because late September in northern Germany reliably delivers near-optimal marathon conditions. The dew point typically sits around 8–10°C, well below the performance threshold. It is no coincidence that 8 of the last 10 men’s marathon world records have been set here.

Boston Marathon — The Wildcard

Average race-day temperature: 13°C. But the average is misleading. Boston’s mid-April date produces extreme variability: historical race-day temperatures range from 3°C (2015) to 27°C (2012, the year of mass heat casualties). In the cool years, Boston is a PR-friendly course despite its hills. In the hot years, it is a survival race. Checking historical weather variance — not just the average — is critical for Boston.

Dubai Marathon — Heat Managed by Early Start

Average temperature at the 6:00 AM start: 19°C. Estimated pace impact: +4.2%. Dubai compensates for its climate with a pre-dawn start, but even 19°C with Gulf humidity (dew point often 15–17°C) creates meaningful heat stress. By the time mid-pack runners reach the halfway mark around 8:30 AM, temperatures have climbed to 22–24°C. Havik’s segment-level analysis captures this intra-race temperature rise, which race-level averages miss entirely.

Tokyo Marathon — Cool but Not Quite Ideal

Average race-day temperature: 9°C. Estimated pace impact: +0.5%. Tokyo’s early March date puts it right at the cool edge of the optimal range. The slight cold penalty is negligible for well-dressed runners, and the low humidity (dew point 2–5°C) means essentially zero humidity impact. It is one of the most underrated weather marathons in the World Marathon Majors.

London Marathon — Consistently Good

Average race-day temperature: 12°C. Estimated pace impact: +0.6%. London’s late April date benefits from the moderating influence of the Gulf Stream. Temperatures are remarkably consistent year to year (10–16°C in 90% of editions), making London one of the most predictable weather marathons. The dew point rarely exceeds 8°C. For runners who want reliable conditions without traveling to Berlin, London is the answer.

Chicago Marathon — October Roulette

Average race-day temperature: 15°C. Estimated pace impact: +1.4%. Chicago’s early October date usually delivers excellent conditions, but it has a long tail of warm outliers. The 2007 edition saw 31°C temperatures that forced race officials to shut down water stations and redirect runners to the finish. In typical years, however, Chicago’s flat course combined with 12–16°C air temperatures makes it one of the fastest marathons in the world.

What Havik Does With This Data

Most marathon weather advice stops at the race level: “It will be 18°C at the start.” That is insufficient. A marathon is not a moment — it is a 3–5 hour journey through changing conditions. Havik models weather at the segment level, and this distinction matters.

Here is how it works:

1. Course segmentation. The race GPX file is parsed and resampled into 1–2 km segments, each with its own latitude, longitude, elevation, and gradient.
2. Arrival-time weather. For each segment, Havik estimates when the runner will arrive based on their target pace, then fetches the weather forecast (or historical average) for that specific time and location. A segment at kilometer 5 (reached at 7:25 AM) gets a different forecast than a segment at kilometer 38 (reached at 10:40 AM).
3. Multi-factor adjustment. Each segment receives a pace adjustment that combines temperature, humidity (dew point), wind (headwind, crosswind, and tailwind decomposed from the wind direction relative to the runner’s bearing), gradient, and altitude. These factors are combined multiplicatively, not additively, because their physiological effects compound.
4. Iterative refinement. After the first pass, Havik recalculates arrival times using the adjusted segment paces and re-fetches weather. This second pass catches feedback loops: if heat slows you down in the first half, you arrive at the second half later, when it may be even hotter.
5. WBGT calculation. Every segment gets a Wet Bulb Globe Temperature estimate using the Stull (2011) approximation, enabling heat stress zone identification that accounts for both temperature and humidity simultaneously.

The result is a per-segment pace plan that reflects actual conditions along the course, not a single number applied uniformly to 42.2 km.

A race that averages 15°C might start at 8°C and finish at 22°C. A single race-level adjustment would underestimate the difficulty of the final 10 km. Havik’s segment model captures this.

Analyze any race on Havik for free →

Practical Advice for Runners

Understanding the temperature-performance relationship is only useful if you act on it. Here are evidence-based strategies for racing in different temperature conditions.

Before Race Day: Choose Your Weather

• Check historical weather before registering. A race’s 10-year weather average is the best predictor of what you will experience. Havik provides historical weather data for every race in its database. If your goal is a PR, prioritize races with average temperatures between 8–14°C and dew points below 12°C.
• Check variance, not just average. A race that averages 14°C but ranges from 5°C to 25°C across years (like Boston) is a riskier bet than one that averages 14°C with a 10–18°C range (like Berlin). Consistency matters.
• Heat acclimate for warm races. If you are racing in conditions above 20°C, invest 10–14 days in heat acclimation before the event. This expands plasma volume, lowers core temperature at a given effort, and improves sweat efficiency. Studies show heat acclimation can recover 50–75% of the heat-related performance loss.

On Race Day: Adjust and Execute

• Adjust your goal pace for actual conditions. If your target is based on a cool training block and race day is 22°C, apply a 3–5% adjustment. Use Havik’s analysis tool to get a segment-specific pace plan. Running the first half at your cool-weather pace in warm conditions is a recipe for a bonk at kilometer 30.
• Hydrate to thirst, do not overdrink. Hyponatremia (low blood sodium from overdrinking) is as dangerous as dehydration. Drink when thirsty. For races above 20°C, aim for 400–800 ml per hour depending on body size and sweat rate. Practice your hydration strategy in training.
• Minimize clothing. Every unnecessary layer adds thermal insulation and impedes evaporative cooling. In temperatures above 15°C, a singlet and shorts is optimal. Arm warmers are a better choice than a long-sleeve top because they can be removed mid-race.
• Start conservative in the heat. The second half of a hot marathon punishes early aggression more than any other racing scenario. The physiological cost of running 10 seconds per kilometer too fast in the first half is amplified by heat stress: higher core temperature, faster glycogen depletion, and earlier onset of cardiac drift. In warm conditions, a negative-split strategy is not just optimal — it is survival.
• Use cooling strategies. Ice in a cap or neck bandana, cold sponges at aid stations, and pouring water over your head (not your shoes) can reduce skin temperature and provide transient relief. Pre-cooling with an ice slurry before the start has been shown to extend time to exhaustion by 10–15 minutes in hot conditions (Ross et al., 2011).

Conclusion

Temperature is the most predictable weather variable and the one with the largest impact on your marathon time. Unlike wind, which changes direction and speed throughout the day, or rain, which may or may not materialize, temperature follows reliable patterns. Historical averages for a given race date and location are remarkably stable year to year.

This predictability is a gift. It means you can choose a race, weeks or months in advance, that gives you the best statistical chance of ideal conditions. It means you can plan your pacing, clothing, and hydration strategy based on data, not guesswork. And it means you can stop blaming bad races on “it was hot” and start accounting for heat in your race plan from the beginning.

The key numbers to remember: optimal is 10–15°C with a dew point below 12°C. Every degree above 20°C costs you progressively more. The curve is exponential, not linear. And the best tool for modeling this is not a static table — it is a segment-level analysis that accounts for how conditions change over the 3–5 hours you are on the course.

Find the best weather marathon for your next PR →

References

1. Ely, M.R., Cheuvront, S.N., Roberts, W.O., & Montain, S.J. (2007). Impact of Weather on Marathon-Running Performance. Medicine & Science in Sports & Exercise, 39(3), 487–493.
2. Febbraio, M.A., Snow, R.J., Hargreaves, M., Stathis, C.G., Martin, I.K., & Carey, M.F. (1994). Muscle metabolism during exercise and heat stress in trained men: effect of acclimation. Journal of Applied Physiology, 76(2), 589–597.
3. Stull, R. (2011). Wet-Bulb Temperature from Relative Humidity and Air Temperature. Journal of Applied Meteorology and Climatology, 50(11), 2267–2269.
4. Ross, M.L.R., Garvican, L.A., Jeacocke, N.A., et al. (2011). Novel Precooling Strategy Enhances Time Trial Cycling in the Heat. Medicine & Science in Sports & Exercise, 43(1), 123–133.
5. Minetti, A.E., Moia, C., Roi, G.S., Susta, D., & Ferretti, G. (2002). Energy cost of walking and running at extreme uphill and downhill slopes. Journal of Applied Physiology, 93(3), 1039–1046.
6. American College of Sports Medicine (2007). Exercise and Fluid Replacement: Position Stand. Medicine & Science in Sports & Exercise, 39(2), 377–390.