For more than a decade the benchmark for human heat survival has been a wet-bulb temperature of 35 degrees Celsius sustained for six hours — researchers have now found that deadly conditions occur well before that threshold is reached

For more than a decade, a wet-bulb temperature of 35 degrees Celsius sustained for six hours has served as a benchmark for the outer edge of human heat survivability in climate science. The figure comes from a 2010 paper by Steven Sherwood and Matthew Huber in the Proceedings of the National Academy of Sciences, which argued this represented the physiological limit beyond which the human body cannot shed heat fast enough to survive without air conditioning. It was intended as an absolute ceiling, not a comfort threshold. But the figure, once established, was widely adopted in climate assessments as a marker for when conditions become deadly, and that adoption has had a specific problem: people have been dying in conditions well short of it.

A paper published on 26 March 2026 in Nature Communications addresses that gap directly. The study, led by Sarah E. Perkins-Kirkpatrick and colleagues including Ollie Jay of the University of Sydney’s Heat and Health Research Centre, applies a newer physiological model called HEAT-Lim to six historical heat events and asks whether the model’s more granular survival thresholds were exceeded during conditions that the Sherwood-Huber 35°C benchmark would have classified as survivable. In all six cases, they were.

Where the 35°C benchmark fell short

The Sherwood-Huber model was built on a straightforward biophysical argument: at a wet-bulb temperature of 35°C, the air is warm and humid enough that sweat evaporating from the skin can no longer cool the body below its core temperature. At that point, without external cooling, core temperature rises until it causes fatal heat stroke. The model does not assume shade, sun exposure, age, or individual variation in sweating capacity. It is a generalised theoretical limit.

HEAT-Lim, developed in a 2023 paper in Nature Communications by Jennifer K. Vanos and colleagues at Arizona State University and the University of Sydney, adds several of those missing variables. It models heat stress as a function of age, access to shade, direct solar radiation, and the specific physiological constraints on sweating, which differ meaningfully between younger adults and people over 65. When the additional variables are incorporated, the thresholds at which conditions become non-survivable are considerably lower and drier than 35°C wet-bulb, and the gap widens further in hot, low-humidity environments and in direct sunlight.

This is the underlying issue the 2026 paper tests: whether HEAT-Lim thresholds were breached during six historical events that produced high mortality but never reached or sustained 35°C wet-bulb temperature.

The six events and what the model found

The events chosen for analysis were: the Middle East heatwave centred on Mecca in June 2024, coinciding with the Hajj pilgrimage; the South Asian heatwave affecting Bangkok in April 2024; the North and Central American heatwave over Phoenix in July 2023; the Australian heatwave over Mount Isa in January 2019; the India-Pakistan event over Larkana in May and June 2015; and the European heatwave over Seville in August 2003. The selection criteria were a combination of climatologically extreme temperatures, high rates of heat-related mortality, or both. None sustained a wet-bulb temperature of 35°C.

The Mecca event showed non-survivable conditions for people over 65 in direct sun for 12 per cent of rolling six-hour periods across the 30-day period studied. The Larkana event reached 18 per cent. In both cases, these findings align with the mortality recorded: deaths among the elderly, in extreme heat, in conditions that the 35°C benchmark would not have flagged as critical.

The dry heat finding

A specific result from the Phoenix analysis is worth noting separately. Hot and dry conditions, where humidity is low but temperature is very high, have sometimes been treated as less dangerous than hot and humid conditions on the grounds that sweat can still evaporate. The HEAT-Lim model incorporates a physiological constraint the Sherwood-Huber model does not: at very low humidity, the human body reaches the upper limit of its sweat production before the environment reaches a wet-bulb temperature anywhere near 35°C. The body simply cannot produce enough sweat to compensate for the radiant and convective heat load. The paper’s authors describe this as an important finding: extremely hot and dry conditions can be just as deadly as hot and humid ones, and this is only beginning to be accounted for in climate risk assessments.

Age and the gap that widens with it

The difference between HEAT-Lim’s thresholds and the 35°C benchmark grows larger for people over 65. Sweating capacity declines with age, and the model incorporates those decrements. In practical terms, this means that during the events studied, conditions were non-survivable for older people in direct sun while remaining survivable for younger adults in the same location. The paper’s figures show that across events in Southern Asia and India-Pakistan, the areas where non-survivable conditions for older people in direct sun overlapped with dense older populations were extensive.

Shade changes the picture substantially. When the model calculates thresholds for older people seeking shade rather than fully exposed, non-survivable conditions drop considerably across all six events. The European 2003 event, which produced over 70,000 excess deaths, shows almost no non-survivable conditions for older people in shade under the model, consistent with the paper’s note that most 2003 deaths in France were cardiovascular or respiratory rather than classic heat stroke. HEAT-Lim models only heat-stroke deaths.

What the paper is arguing for

The authors are arguing for a methodological update: that climate risk assessments, which have used the 35°C wet-bulb figure as an indicator of when conditions become lethal, should shift to physiology-derived models that account for age, shade access, solar radiation, and individual sweating capacity. The implications are that current assessments have been underestimating the geographic footprint of already-deadly heat, particularly for older populations in low-latitude regions.

This is one study applying one model to six case studies. The HEAT-Lim model itself was published in 2023 and is still being tested and refined. The six events were chosen partly because they were climatologically extreme or associated with high mortality, which means they are not a random sample of heat events. What the paper establishes is that the gap between the widely-used benchmark and actual physiological limits is real, measurable, and has been large enough to matter in six significant historical cases.