Who a Phoenix summer hits hardest

A Phoenix summer is usually described as a temperature. It is more usefully described as a set of overlapping seasons (the run of days above 100 °F, the run of nights that never fall below 80 °F, and the months the county opens its heat relief sites) that start and end on different dates and leave the shoulders of the year in a gap.

The heat gets its shape from the climate record. Who it harms gets its shape from a different dataset entirely, and the two do not line up. That mismatch is the subject of this page: the season as the normals describe it, and the deaths as Maricopa County recorded them in 2024.

Everything here is a published figure with a source beside it. Health and safety pages on this site cite official sources only and give no individual medical advice; where a decision turns on a person’s own health or medication, the answer is a clinician, and where somebody is confused, fainting or fitting, the answer is 911.

Reading the overlap

The windows nest rather than stack. The county’s heat relief sites open on 1 May, thirty days before the normal high reaches 100 °F, and close on 30 September, twelve days after it drops back below. The 100 °F window runs 31 May to 18 September: 111 days. Inside it sits the 105 °F window, 18 June to 22 August, 66 days, which is the stretch most likely to produce an alert.

The two night windows are the ones people underestimate. The normal low sits at or above 80 °F for 84 nights, from 20 June to 11 September. That is the run that makes a house without working air conditioning dangerous at 3 a.m. rather than at 3 p.m., and it begins three weeks before the 105 °F days and ends three weeks after they do.

Two of the five rows measure the night, and for a reader deciding whether a fan will do, they are the rows that matter. July’s normal low is 84.5 °F and its normal high is 106.5 °F, so the day’s relief is a fall from very hot to merely hot.

What the county’s record adds

The climate record says how long the season runs. It says nothing about who it kills, and for that the only dataset close to home is Maricopa County’s annual heat-death report.

Its 2024 report recorded 602 confirmed heat-related deaths, with eight still under investigation, against a record 645 in 2023, the first year-over-year decline since 2014, and a decline in the same year the county describes as the hottest summer on record. From 18 June to 31 July, at least one heat death was recorded every day, a run of 44 days.

Four figures from that report are worth holding on to:

• 51% of the year’s heat deaths happened on days the heat-risk forecast rated only moderate, a level below the one that triggers an alert.
• Among indoor deaths, 88% had an air conditioning unit present, and 70% of those units were not working. The unit that fails is a bigger factor than the unit that is absent.
• 349 deaths involved substances, and 89% of those involved stimulants. The county names substance use among its own risk factors.
• The groups most affected, in the county’s words, were people experiencing homelessness, older adults, and people who use substances.

The report was preliminary and the county later replaced it with a final version whose totals this site has not opened, so the counts here carry that date. The pattern does not depend on the last few investigations either way.

Who the agencies name

The climate figures describe a season. The risk lists describe a person. The CDC and Maricopa County answer the second question differently, and the differences are worth seeing side by side.

Read together, the two lists are less a contradiction than a division of labour. The CDC’s list is built for anyone, anywhere, and is dominated by the heat a person meets outdoors and by age. The county’s list is built for this county, where the deaths it recorded inside homes make housing, cooling and medication the operative words. Neither list is wrong. A reader who is in only one of them is still in one of them.

Where the two datasets disagree

Three mismatches between the season and the deaths are worth naming rather than smoothing over.

The first is the moderate days. The season is defined by temperature, and the warnings are defined by a forecast, but half of the 2024 deaths fell on days the forecast judged only moderate. A day that does not trigger an alert is not a day without risk, and the county’s own record is the evidence.

The second is the indoors. The county’s figures point at cooling that exists but has stopped, in homes where the occupant may not have noticed until the night was already hot. The 84-night run above 80 °F is exactly the window in which that failure matters and in which the outdoor temperature is not extreme enough to warn anyone.

The third is the night itself. Almost every measure a person checks on a summer morning is a daytime measure, and the warm-night window opens three weeks before the hottest days do. The heat a person sleeps through is the heat that has no alert attached to it.

The two seasons do not line up

Set the relief season against the temperature season and the shoulders are obvious. The county opens its sites on 1 May, a month before the normal high reaches 100 °F, and closes them on 30 September, after the 100 °F window has ended but before the warm nights have fully gone.

The gap that remains is the mild-night window, which runs to 5 October. Nights that stay above 70 °F are not a heat emergency, and they are also not the point at which a body recovers well. The relief season is sized for the days that make the news; the tails belong to the temperatures that do not.

What follows from that

For a household, the list is short. Know which of the two agency lists you are on, and treat the night as part of the day: the Heat-Safe Hours Planner prints the whole 24 hours for a month, not just the daylight. Check that the cooling works before the run of 80 °F nights, not during it. Check on the people the county names: older neighbours, anyone living outside, anyone whose medication affects how they handle heat.

Nobody should be diagnosing heat illness from an article. The Heat-Illness Triage Guide works through heat exhaustion and heat stroke symptom by symptom, in the CDC’s own wording, and the comparison of the two conditions explains why the confusion between them is the dangerous part. For water and salt, what the health agencies publish is the place to start.

If somebody is confused, has slurred speech, has fainted or is fitting, has skin that is very hot, or has stopped sweating (or is sweating heavily with hot skin) call 911. Heat stroke is a medical emergency, and the first step is the phone.