The Hajj pilgrimage to Mecca is one of the five pillars of Islam and a lifelong dream for many Muslims. All Muslims who are physically and financially able are expected to make the pilgrimage at least once in their lives. This year, the Hajj falls in May, traditionally a safer and milder time of year than the scorching summer months of June to September, and 2 million people are expected to embark on the pilgrimage (Al Jazeera, 2026). The pilgrims often undertake walking long distances on a daily basis, making it much harder under extreme heat conditions.

The timing of the Hajj is determined by the Islamic lunar calendar. Hajj takes place every year during the month of Dhu al-Hijjah, the twelfth and final month of the Islamic calendar, with the main rituals occurring between the 8th and 13th days of that month. Because the Islamic calendar is based on the moon, it is about 10–11 days shorter than the solar Gregorian calendar used in most of the world. As a result, the dates of Hajj shift earlier each year relative to the seasons. Over a cycle of roughly 33 years, Hajj moves through every season from the cooler winter months to the extreme heat of summer (Yezli et al., 2024).

But climate change has altered those patterns. Temperatures in Saudi Arabia are rising earlier in the year and reaching more dangerous extremes, making the pilgrimage increasingly hazardous not only in the summer months, where in 2024 more than 1300 pilgrims lost their lives when Hajj started in mid-June temperatures reached 51°C (The Conversation, 2024), but even during the spring months (Figure 1).

In a super-rapid study researchers from World Weather Attribution show that for pilgrims who spend 20–30 hours outdoors, often walking in dense crowds, the risk of heat exhaustion and heat stroke is becoming far more severe earlier in the year, with temperatures now reaching heights in May that were only seen in June to August in the 1960s and 1970s, due to human-induced climate change.

Figure 1: (a) Estimated local increase in May mean temperatures across Saudi Arabia, compared to a 1.3°C cooler climate. Mecca (39.8E, 21.4N) is marked with a cross. Box shows the study region, a 2° square box centred on Mecca. (b) Monthly mean temperatures over land surface within the study region. Transparent lines show monthly maximum or mean per calendar month; bold lines show a 10-year rolling mean. Dashed line shows the 10-year mean from 2016-2025; May mean temperatures have now reached levels that, prior to 1980, would have only been experienced during the summer months. All data: ERA5.

Key Messages

Saudi Arabia’s climate has always been extremely hot, and Hajj pilgrimages that fall during the summer months have long carried risks of heat stroke and other heat-related illnesses. As temperatures rise further due to climate change, those dangers are becoming more severe. In response, Saudi authorities have introduced extensive heat action plans, including shaded walkways, cooling stations, misting systems, and expanded medical services, which have helped reduce cases of heat stroke and other heat-related illnesses among pilgrims (Yezli et al., 2024).
When analysing gridded observational-data products, we find that the average daily temperatures during May 2026 are almost as high as the June-August temperatures in the period 1970 to 1990, reaching around 31.2°C in Mecca. This means what used to be a significantly cooler and thus safer time of year for Hajj is now as dangerous as the height of summer used to be.
We find that daily maximum May temperatures of 40°C, historically more typical of the maximum temperatures seen during the peak summer months of June, July, and August, have become far more common compared with a world that was 1.3°C cooler. Such temperatures are now expected to occur in May every two to three years. Peak May temperatures are also now approximately 2°C hotter.
We also find that average May temperatures are now approximately 3.5°C higher than would have been in a 1.3°C cooler climate. While this year’s mean May temperature of 31.2°C is fairly commonplace in today’s climate, these temperatures are now many times more likely than they would have been in a preindustrial climate. May mean temperatures above 32°C – the mean summer temperature from 1970-1990 – are now expected to occur most years, highlighting how climate change is shifting extreme heat earlier into the year.
While Saudi authorities have implemented a range of heat mitigation measures to reduce heat-related risks during Hajj, access to these protections has not been uniform. In particular, pilgrims without official permits may face limited access to safe food, water, cooling infrastructure and medical support. Strengthening and expanding equitable access to heat protection measures will be essential to ensuring the safety of all pilgrims, especially those most vulnerable to extreme heat exposure.
Research has shown that this trend will continue to intensify as the planet warms. One study found that if global temperatures rise by 3°C by the end of the century, roughly the trajectory current climate policies are putting the world on, around 97% of all Hajj pilgrimages would take place during periods when dangerous levels of heat are expected in Mecca. Thus, a rapid transition away from fossil fuels is essential to avoid ever more dangerous Hajj.

Known trends in heatwaves in the Arabian Peninsula

IPCC AR6 synthesises that there is medium confidence that human-induced climate change is increasing hot extremes in the Arabian Peninsula (IPCC, 2021). This is a region that has experienced accelerating warming, increasing from about 0.10°C per decade during 1901–2010 (Attada et al., 2018) to around 0.63°C per decade since 1978 (Almazroui, 2020). Projections of annual temperatures in the region show expected increases 1.6°C under low-emissions scenarios (SSP1-2.6) and up to 5.3°C under very high emissions (SSP5-8.5) by 2070–2099, as compared to 1980-2010, with the greatest warming expected in northern areas (Almazroui, 2020).

In 2024, the World Weather Attribution (WWA) initiative carried out a rapid attribution study to quantify the role of climate change in the extreme heatwave that affected large parts of Asia that year (Zachariah et al., 2024). Although the study region focused on a region over Israel, Palestine, Syria, and Jordan, and lies to the north of the current study area, the nature of the trends is consistent with findings for the wider Arabian Peninsula and surrounding regions. The study concluded that human-induced climate change made such extreme heat events substantially more likely and more intense – around five times more likely and approximately 1.7°C warmer compared to pre-industrial climate.

Analysis of trends in extremes

Event definition

Warming in recent decades means that May temperatures now regularly reach levels previously more typically experienced during the summer months (Figure 2). In this short observations-only analysis we examine trends in both monthly mean temperatures and maximum 1-day temperatures during May (denoted, respectively, Tg-May and Tx1x), over the land surface within a 2-degree box centred on Mecca (39.8E, 21.4N). Due to the unavailability of long time series of station-based observations in this area, we predominantly use ERA reanalysis from 1950-2025 (Hersbach et al., 2020), extended with analysis data from May 1st-25th and forecast data until the end of the month. We also consider trends in daily maximum temperatures from NOAA’s CPC dataset; this dataset only begins in 1979 and so uncertainties are typically much higher.

Figure 2: Seasonal cycle of daily mean, maximum and minimum temperatures over the study region outlined in Figure 1 (ERA5). Red lines show the 1990-2020 mean, blue lines show the 1950-1980 mean.

The methods used to analyse heat trends follow the standard WWA protocol using non-stationary extreme value theory, as described in Philip et al. 2020 and expanded upon in Otto et al., 2024. The peak temperatures (Tx1x) are modelled using a generalised extreme value (GEV) distribution, while monthly mean temperatures (Tg-May) are modelled with a normal distribution; both indices are assumed to increase linearly with global mean surface temperature. A more detailed description of this method and an example can be found in Clarke et al., 2026.

While we present changes in the likelihood and intensity of events with respect to the preindustrial climate (1850-1900, based on the Global Warming Index), we note that almost all of the observed warming has occurred since 1970, meaning that events that would have been rare in a preindustrial climate would have been almost as rare during that period.

Results

Trends in peak May temperatures

Both the ERA5 and CPC datasets show a clear increase in May peak temperatures (Tx1x) over recent decades (Figures 3 and 4); in both datasets, peak May temperatures are around two degrees hotter than they would have been in a preindustrial climate (Table 1), although in the shorter CPC dataset, the range of uncertainty is high. At the time of writing, temperatures in Mecca were forecast to exceed 40°C throughout the last week of May; in ERA5 forecast and reanalysis data, the maximum temperature across the region was 40.9°C, and in CPC 41.9°C. These temperatures are not unusual for May in the current climate, with a return period of two to three years in both datasets. However, such extreme heat would have been much less likely in a preindustrial climate: in ERA5, similarly hot temperatures are around 25 times more likely due to climate change, and in CPC, around 13 times more likely. Similar trends are observed in the warmest nighttime temperatures (not shown).

The high temperatures seen in May 2026 – which are actually fairly typical in today’s climate – fall within the ranges of annual maximum temperatures recorded between 1970 and 1990, all of which occurred during the summer months from June to August. In ERA5, peak summer temperatures ranged between 38.1°C and 41.2°C, while in CPC (which begins in 1979), peak temperatures prior to 1990 ranged from 40.5°C to 43.9°C. Dangerously hot temperatures that, prior to 1990, would have been expected only during the summer peak are now routinely occurring during May, meaning that the period of dangerous heat is much longer than it would have been previously (Figure 1b).

Figure 3: (left, centre) Peak May temperatures (Tx1x) in the region surrounding Mecca, with fitted trend overlaid. Solid black line denotes the expected temperature on the hottest day each May; blue lines indicate the expected magnitude each year of a 1-in-6-year (thick line) and 1-in-40-year temperature event (thin line). Green dashed line is a nonparametric Loess smoother.
(right) Expected return levels of Tx1x according to the fitted statistical model: in the 2026 climate (red line) and in a 1.3°C cooler counterfactual climate (blue line). Shaded regions represent 95% confidence intervals obtained via a bootstrapping procedure. The pink line shows the daily maximum temperature over the study region. Red and blue ticks at the x axis indicate the estimated return level of the event in the 2026 and counterfactual climates.
All data: ERA5.

Figure 4: As Figure 2, but for the CPC dataset.

Table 1: Results of Tx1x trend fitting: maximum daily temperature (averaged over the study region) during May 2026; estimated return period of that temperature in the current climate; change in likelihood and intensity of recording similarly extreme temperatures with respect to a 1.3°C cooler preindustrial climate; mean and range of summer maximum temperatures from 1970-1990 (*1979-1990 for CPC).

Trends in mean May temperatures

May mean temperatures exhibit an even stronger warming trend (Figure 5); in ERA5, average May temperatures are around 3.5°C warmer in 2026 than they would have been in a preindustrial climate (Table 2). At the time of writing, the average temperature in May 2026 was expected to be 31.2°C, although this includes five days of forecast data, so final reanalysis values may differ slightly. Once again, these temperatures are fairly typical in the current climate, with a return period of around one year, meaning that May is almost always expected to be as warm as this, or warmer. However, due to the strong warming trend, similar temperatures would have been very rare in a preindustrial climate, with the estimated probability ratio over 10,000. May 2026 – a fairly moderate May in today’s climate – is as warm as the average summer (June-August) between 1970 and 1990.

Figure 5: As figure 2, but for May mean temperatures (ERA5).

Table 2: Results of Tg-May trend fitting: mean May temperature (averaged over the study region) in 2026; estimated return period of that temperature in the current climate; change in likelihood and intensity of recording similarly extreme temperatures with respect to a 1.3°C cooler preindustrial climate; mean and range of mean summer temperatures from 1970-1990.

Study Authors:

Clair Barnes, Centre for Environmental Policy, Imperial College, London UK
Friederike Otto, Centre for Environmental Policy, Imperial College, London, UK

Review Authors:

Mariam Zachariah, Centre for Environmental Policy, Imperial College, London, UK
Sjoukje Philip, Royal Netherlands Meteorological Institute (KNMI), De Bilt, The Netherlands
Emmanuel Raju, Department of Public Health, Global Health Section and Copenhagen Centre for Disaster Research, University of Copenhagen

This study reflects the views of the listed authors.