What Paris's heatwave playbook teaches cities that didn't write one

On 24 June 2026, the thermometer in Paris read 33°C by early afternoon, and the city's prefectoral authorities moved into the first stage of a plan that did not exist twenty-three years ago. France's national heat alert system, written in the weeks after an August 2003 heatwave that killed roughly 15,000 people, is now the operational backbone that decides whether parks stay open after sundown, whether public drinking fountains run at full pressure, and whether building sites are allowed to pour concrete. It is, in effect, a working draft of what climate adaptation looks like when a capital city is forced to write one.

The story behind the system, and the question of what other cities can borrow from it, is worth pausing on. Heat is the most democractic of climate hazards: it touches rich and poor neighbourhoods alike, but it kills unevenly, and the institutions built to anticipate it reveal more about a city's capacity to govern than almost any other policy. Paris's answer was not a piece of legislation but an operational doctrine — colour-coded alert levels, public behaviour attached to each, and the legal authority to compel employers and property owners to act. Few cities outside France have anything approaching it.

The architecture: four levels, named obligations

The French system, run by Météo-France and the health ministry, sorts heat risk into four colour-coded tiers: green (seasonal vigilance), yellow (heatwave), orange (extreme heatwave), and red (exceptional heatwave). Each level unlocks a fixed menu of obligations. At yellow, the state activates a public information line and reminds employers of their duties; at orange, school outings are reviewed and nursing homes are checked on residents with disabilities; at red, public events can be cancelled by prefectoral order and non-essential outdoor work can be suspended.

What makes the French architecture distinctive is not the colour scale — meteorologists use similar tools across Europe — but the chain of obligations attached to each tier. Cities are not asked to "consider" measures; they are told which measures become compulsory and given the prefectoral authority to enforce them. The system was last fully exercised in 2024, when a stretch of late-summer heat pushed several departments into orange and parts of southern France briefly into red. According to Helen Massy-Beresford's reporting for The Guardian, the playbook has since become a quiet reference point for mayors elsewhere in Europe who find themselves improvising their first heatwave.

What the 2003 disaster changed

The August 2003 heatwave was, in retrospect, a governance failure as much as a meteorological one. France lost roughly 15,000 people in two weeks, with the heaviest toll among elderly residents of un-air-conditioned apartments and patients in understaffed care homes. The political reckoning produced the current system within a year: a permanent inter-ministerial crisis cell, a national heat observatory, and the legal basis for prefectoral orders that can override local discretion during declared emergencies.

The deeper legacy is cultural. French public-health messaging now treats heatwaves as expected, recurring events rather than exceptional ones. Summer planning in Paris, Lyon and Marseille begins in May, with city services briefed on which buildings in their districts house vulnerable residents and which public spaces can be opened overnight. The list of cooling sites — libraries, museums, mairie annexes — is published in advance. That kind of pre-positioning is what most non-French cities still treat as improvisation.

The counter-case: cities that improvised

The cities that have received the most attention for heat response in recent years — Phoenix, Athens, Seville, several Indian metros — have done so without a comparable national framework. Seville, working with local universities and the regional Junta de Andalucía, has produced its own naming system for heatwaves and a tiered municipal response that mirrors parts of the French model without the prefectoral authority behind it. Phoenix's cooling centres are run through a public-private partnership with the city's utilities; Athens has leaned on the Acropolis's closing hours as a de facto behavioural lever.

The point of comparison is instructive. Cities without a national doctrine spend the first 72 hours of a heatwave arguing about whose budget pays for bottled water and which agency is allowed to open a school gymnasium. Cities operating inside the French doctrine have already rehearsed those decisions in tabletop exercises and have pre-signed memoranda with hospitals, transit operators and large employers. The marginal cost of the system, in steady state, is a small inter-ministerial cell and a public-health communications budget. The marginal cost of not having one was visible in 2003 and has been visible in lower-casualty form in cities that have improvised poorly since.

The structural frame: heat as a planning problem, not a weather problem

What Paris is selling, implicitly, is the proposition that heat is a planning problem disguised as a weather problem. A forecast is a one-week-ahead data point; a city's exposure to that forecast is the product of building stock, tree canopy, the density of un-air-conditioned rental housing, the operating hours of public pools, and the legal authority to compel cooling measures inside private workplaces. Once heat is reframed that way, the relevant policy levers are not meteorological but municipal: building codes, parks budgets, labour inspectors, the discretion granted to a prefect.

The wider pattern is consistent with what other adaptation debates have produced. Cities that treated flooding, air quality, or wildfire smoke as discretionary concerns handled by parks departments ended up improvising. Cities that wrote those hazards into the same operational doctrine that handles snowstorms and major fires ended up with response times measured in hours rather than days. France's heatwave system is, in that sense, a case study in how a state converts a recurring shock into a routine.

Stakes: what a hotter decade will demand

The trajectory is not ambiguous. European summers in 2024 and 2025 both broke records in regions unaccustomed to sustained heat, and Mediterranean cities that once planned around July now plan around late May. The question for non-French mayors is whether to copy the French architecture in full — colour codes, prefectoral authority, fixed obligations — or to assemble a lighter-weight version that fits their own legal order. Either path is cheaper than the alternative, which is to repeat the 2003 pattern of death in nursing homes while officials debated whether the temperature qualified as a crisis.

The harder, quieter question is who pays for the cooling. Public pools, shaded tram stops, retrofitted social-housing blocks — these are capital expenditures that show up years before they save lives, and they compete for budgets that are already tight. Cities that have begun the work treat it as infrastructure. Cities that have not treat heatwaves as weather. The difference, in a 40°C August, is measured in body bags.

What remains uncertain

The available reporting does not specify how the French system's effectiveness has been independently audited since 2003, nor whether the orange and red tiers have been triggered often enough to test the prefectoral orders in court. Cities that have studied the French model — Seville is the most-cited example — report favourable results, but the comparative data are thin. What the sources do show, unambiguously, is that the French state wrote the playbook after a disaster it did not anticipate, and that the playbook has held through twenty-three subsequent summers. Other cities now have the chance to write theirs before theirs.

Desk note: The Guardian's reporting, summarised above, frames France's response as a public-health success story; this publication treats it as a planning case study — what an operational doctrine for a recurring climate shock looks like, and what the absence of one costs.