How TownRatings grades French towns

Every grade in the French edition comes from public data, compares a commune only with other French communes, and is turned into a letter the same way. This page explains how, once, so each town page can stay about the town. American and British towns are graded separately, in their own editions.

From numbers to letters

A town is a commune, the French town with its own mayor (INSEE’s list of 1 January 2026); Paris, Lyon and Marseille are split into their 45 arrondissements, as London is split into boroughs. For each subject we rank every French commune of 1,000 or more residents on the public data behind it. A commune’s percentile is the share of those communes it does better than. Where a subject has several measures, each is ranked, the ranks are averaged, and the average is ranked again. The percentile becomes a letter on one fixed scale:

Bottom 5% · then steps of 10% · top 5%. There is no F: a D− town is still a real place where people live and make it work.

The letters are not marks out of 20. A 14/20 says how well a pupil did against a standard; our letter says where a town stands among French towns. An A means better than about 85% of them, a C sits in the middle, and a D is in the bottom fifth. The same share of towns gets each letter in every subject.

Each town also shows its rank within its département, because the national letter and the local picture can differ. A département with fewer than five graded towns gives no département rank; its towns show their rank in France.

Which towns get grades

• 2,500 residents or more: a grade for each subject the data allows, an overall grade, and ranks in France and in the département.
• 1,000 to 2,499 residents: subject grades, but no overall grade and no rank. The figures are thinner in small communes, so a single subject is as far as we go.
• Under 1,000 residents: a page with the commune’s figures, and no grades. Most of France’s 34,858 communes are this size; a handful of crimes or deaths can swing their figures from year to year.

Residents means one figure everywhere, the same one every page shows: INSEE’s municipal population (population municipale) from the census in force since 1 January 2026.

The overall grade

Six subjects are graded. The overall grade blends them by weight: cost of living 18, safety 18, schools 16, health 14, jobs and income 12, walkability 8. The blend is ranked again across France, so an A means the same share of towns everywhere. A town needs 2,500 residents, a Safety Grade, at least three of cost of living, safety, schools and health, and at least five subjects in all. A subject a town has no grade for is left out of its blend, and the others’ weights are scaled up to fill it. Flood risk has a weight in our other editions; in France it isn’t graded (see below), so it is left out the same way.

Because Flood risk isn’t graded, cost of living and safety carry a little more of the overall grade than in the U.S. and UK. Big cities, where both are harder, read a little harsher as a result.

Safety

Crimes recorded by the police and gendarmerie in the commune, from SSMSI’s figures by commune, for 2023, 2024 and 2025, per 1,000 residents a year (each year’s INSEE population). The grade counts eight kinds of crime against residents and their homes: violence within the family, other physical violence, sexual violence, armed robbery, burglary, vehicle theft, theft from vehicles, and criminal damage. Violent crime counts once, as property crime does, the UK’s rule (the U.S. weighs it four times). The counts mix units (victims, offences, vehicles), so the result is a weighted index, never a number of crimes or a resident’s chance of being a victim. Drug offences, fraud and theft without violence are shown on each Safety page but not graded.

SSMSI hides a commune’s count for a crime when it had five or fewer cases in one of the years, to protect victims. A hidden count is never treated as zero. In its place we use SSMSI’s own figure, the average for the département’s communes whose count is hidden. That average belongs to the département, not the town, so we measure how much of each town’s Safety index rests on it (most often for armed robbery, a rare crime):

• 10% or less: graded as usual.
• Over 10%, up to 25%: graded, and the page says that some rarely recorded crimes use the département’s average where the town’s count is hidden.
• Over 25%: not graded, with that reason. A grade is never mostly borrowed. This mostly affects small places, where rare crimes are most often hidden: most communes of 1,000 to 2,499 have no Safety Grade for this reason, and nearly all the larger towns without one have fewer than 5,000 residents.

Each Safety page shows a “how sure” range: the town’s place if its recorded counts moved by chance, plus the whole of any borrowed part either way. Places that draw visitors and shoppers record more crime per resident than they would per person present; their pages say so.

Source: SSMSI, crime recorded by the police and gendarmerie by commune (base communale), 2023 to 2025; INSEE populations.

Schools

The collèges and lycées within 5 km of the town, weighted by candidates. Collèges on the brevet (the pass rate) and their value added (IVAC); lycées on the bac pass rate and their value added (IVAL), general and technological lycées and vocational lycées each against their own kind. Value added compares a school’s results with what its pupils’ backgrounds and earlier results predict, so a school isn’t marked up just for a privileged intake. Each result is set against French schools of the same kind, pass rate and value added are averaged, and the town’s schools are combined. A town needs a collège with published results within 5 km; lycées count where there are any.

Primary schools aren’t graded: national test results aren’t published school by school. The social-position index (IPS) is shown on each page as context, never graded.

Source: Ministry of Education (DEPP): brevet and bac results 2025 with IVAC and IVAL value added; directory of schools; IPS 2024–2025.

Health

One measure, the town’s death ratio over ten years: deaths of residents from 2016 to 2025 against the deaths France’s death rates for each age and sex would give a town with these residents. A ratio of 1 is France’s average; lower is better. Allowing for age matters: a town of retired people isn’t marked down for being old.

One step is ours, not INSEE’s. Deaths are counted where people lived, so a town with a large care home (Ehpad) records the deaths of very old people who moved there from elsewhere. We divide each town’s ratio by a national care-home factor (deaths rise by about 0.66% for each care-home place per 100 residents aged 80 or over, estimated across all French communes). This adjustment is a TownRatings modelling step, not an official statistic.

Small towns have few deaths, so their ratio can move by chance. Each Health page shows a 90% range for the town’s place, from 1,000 simulated redraws of its deaths. Where that range is wider than 40 points, the town is still graded, with a “less sure” note.

Access to care (how many GPs, nurses, physiotherapists, midwives and dentists are within reach, the nearest emergency department, maternity unit and hospital), heart and stroke figures for the town’s area and the département’s life expectancy are shown on each Health page but not graded: they describe care, or a wider area, not the residents’ own health.

Sources: INSEE deaths by commune of residence, 2016–2025; INSEE census populations; Eurostat death rates for France; INSEE equipment base (BPE 2025, care-home places); DREES (APL care access); Santé publique France (Odissé); INSEE life expectancy by département.

Cost of living

Two measures, each against the median income of the town’s area: what a typical 70 m² home costs (the median price per m² of houses and flats sold over three years, where the town had 30 or more sales), and a year’s predicted rent for a 70 m² flat. The area is the town’s intercommunality (EPCI), France’s nearest equivalent to a council area: area income is used, not the town’s own, so a wealthy town isn’t graded as cheap just because its residents earn a lot. Pages show both, labelled. The four island communes that belong to no intercommunality (Île-de-Bréhat, Île-de-Sein, Ouessant and L’Île-d’Yeu) use their own income.

Rent is required; the price counts where it is published. Alsace and Moselle keep their own land register, so no sale prices are published there: towns in Bas-Rhin, Haut-Rhin and Moselle are graded on rent alone, and their pages say why. Rents are the national rent map’s predicted advertised rents, charges included, not rents people actually pay.

The property tax rate (taxe foncière) is shown on each page but not graded: it is a rate on a cadastral value, not a bill.

Sources: DGFiP DVF (geolocated, Etalab), sales July 2023 to June 2026; ANIL/DHUP rent map 2025; INSEE Filosofi 2023 (median disposable income per consumption unit, by intercommunality); DGFiP REI.

Jobs and income

Three measures: the town’s median disposable income per consumption unit (INSEE’s measure of living standards, which counts household size; it is not household income); unemployment among 15 to 64-year-olds at the census; and jobs in the town per employed resident. Two of the three are needed. Where INSEE doesn’t publish a small town’s income, the other two are used.

Sources: INSEE Filosofi 2023; INSEE census 2023 (employment, jobs at place of work).

Walkability

Two measures together: the share of residents living within 800 metres, in a straight line, of each of six everyday places (a bakery, a pharmacy, a grocery, a primary school, a GP and a post office), averaged; and the share of commuters who walk, cycle or take public transport to work. Where residents live comes from INSEE’s 200-metre population grid, so a commune’s empty land doesn’t count against it.

A straight line is a shortcut: a river or a motorway can make 800 metres a longer walk. Towns without the population grid (Guadeloupe and French Guiana) have no Walkability Grade.

Sources: INSEE equipment base (BPE 2025, geolocated); INSEE Filosofi 2021 200 m grid; INSEE census 2023, means of transport to work.

Flood and hazards: shown, not graded

Flood risk is graded in our other editions, from national flood maps. France has none yet: flood maps cover designated risk areas (TRI) only, so a town outside them would look dry when it simply hasn’t been mapped. The history of flood disaster declarations (CatNat) doesn’t measure exposure either. So no French town has a flood grade, and none is counted in the overall grade.

Each town’s Flood & hazards page shows everything that is published: disaster declarations by peril and the last flood, risk prevention plans, clay shrink-swell, the seismic zone, radon, mapped flood zones where they exist, and industrial (Seveso) sites. We will grade flood risk if a national flood map is published.

History and the outdoors

History and the outdoors describe a town’s character. In our other editions each has a grade on its own page that never counts toward the overall grade. In France both are described, not graded, for now: the History page lists the town’s classé and inscrit historic monuments, and the Outdoors page its parks, forests, coast and paths.

Climate and weather are described, not graded, everywhere: weather isn’t better or worse, only a matter of taste.

Guadeloupe, Martinique, French Guiana and Réunion

Guadeloupe, Martinique, French Guiana and Réunion are graded with the rest of France, on the same scale. Mayotte isn’t covered yet: it isn’t in the census file the pages use, and it has no sale prices.

INSEE publishes no Filosofi income figures for Guadeloupe, Martinique and French Guiana, and cost of living is always measured against income. So their towns have no Cost of Living Grade and, as a result, no overall grade; they keep their grades for safety, schools, health and jobs, and walkability in Martinique. Réunion has every figure and is graded in full.

Towns without a grade

• Towns under 2,500 residents get subject grades but no overall grade; under 1,000, no grades at all (see above).
• No Safety Grade: where more than a quarter of the town’s Safety index would rest on the département’s average for hidden counts. Without Safety there is no overall grade.
• No School Grade: where no collège with published results is within 5 km.
• No Cost of Living Grade: in Guadeloupe, Martinique and French Guiana (no income figures), or where no rent is published.
• Paris, Lyon and Marseille as a whole: the city pages gather their arrondissements; each arrondissement is graded and ranked as a town.

What the grades can’t tell you

• Bigger places record more crime per resident. Town centres draw people from far beyond their residents, so cities grade lower on Safety than villages almost everywhere.
• Official data, as published. We use figures as their agencies publish them and say where each comes from; where a figure is hidden, missing or modelled, the page says so.
• How sure we are depends on the town. Small towns have small counts, so their grades can move with a few crimes or deaths. Safety and Health pages show the likely range.
• An average isn’t a street. A town’s grade averages neighbourhoods that can differ more than whole towns do.
• Grades refresh when the data does. Each page says which years its figures cover.

Contains public sector information under the Licence Ouverte / Open Licence 2.0 (Etalab): INSEE, DGFiP, the Interior Ministry’s statistics service (SSMSI), the Ministry of Education (DEPP), the Health Ministry’s statistics service (DREES) and the Housing Ministry (Carte des loyers). Commune boundaries: IGN Admin Express.