# How many auto repair shops are there really in France?

> The full census of auto repair and body shops in France: 101,989 locations, of which 60,059 at the core of the trade and 13,292 hidden franchised agents.

Source : https://scalon.fr/en/observatory/car-repair-shops

---

[← The Observatory](/en/observatory)

Observatory · 2026 census

The industry study says 25,000. The registry counts 107,000. We counted 101,989, of which 13,292 that no export ever returns.

Mainland France Surveyed July 2026 96 districts 8-page PDF report

The short answer

Mainland France has **60,059** repair shops in the strict sense, meaning mechanical and body shops reachable by phone, and **101,989** once franchised agents, inspection centers and roadside assistance are added. Both figures are correct. They do not answer the same question.

Published 4 August 2026 · Next update July 2027

What the survey says

101,989

Full market

Every trade in automotive aftersales, franchised agents included.

60,059

Core of the trade

Mechanical and body work only, with a phone that rings.

13,292

Invisible to exports

Shops filed as dealerships by automatic categories.

6,512

Ruled out by hand

Examined one by one, then taken out of scope, with a reason.

## 01 · Why published figures run from 25,000 to 108,000

An industry study says 25,000 repair shops. An export by activity code returns more than 100,000. Online maps show around 96,000. In a market this size, a fourfold gap is not measurement noise, it is a disagreement about definitions that nobody has bothered to settle in public.

The word "repair shop" covers at least five different trades. A general mechanical shop, a body shop, a tire shop, an inspection center and a tow operator do not sell the same things, do not buy the same things, and are not interchangeable prospects. Counting them together or separately changes the result by a factor of two.

Our position: there is no single figure, there are three, and the only useful work is saying which one answers your question.

### Who publishes what, and on what scope

The five counts below are all correct. None is comparable to another, because none counts the same thing: a collectively bargained branch, a statistical sector, an activity code, a network of banners, and a recompilation of ranges. The fourfold gap is not a measurement error, it is five different questions that were given the same name.

| Source | Published figure | What it actually counts | Vintage | Breakdown available |
| --- | --- | --- | --- | --- |
| [ANFA, French automotive services observatory](https://www.anfa-auto.fr/observatoire/les-secteurs/commerce-et-reparation-automobile) | 128,000 | The companies in the collectively bargained "automotive sales and repair" branch. A labor-law scope, not an activity classification: new vehicle sales is included in it. | 2023 data | No breakdown published by shop trade. The 336,350 employees are split by job function, not by type of business. |
| [INSEE, sector fact sheet 452](https://www.insee.fr/fr/statistiques/7929133?sommaire=7763898) | 57,283 | Sector 452 "Maintenance and repair of motor vehicles", within the national statistical survey. Companies, not locations: that is where most of the gap with the other counts comes from. | 2021, Esane survey | Headcount, revenue and margin brackets. Nothing on the type of shop or on network membership. |
| An export by activity code 4520A | ≈ 65,000 locations | The locations carrying the code, head offices included, with no filter on actual activity or on closure. What an off-the-shelf purchased file returns. | Continuous feed, never dated | The code, and nothing else. A shop registered under retail does not appear; a head office with no shop does. |
| [Independent repair networks, trade press](https://j2rauto.com/reseaux/mra-la-riposte-des-independants-face-aux-reseaux-constructeurs/) | 21,000 to 24,750 | Shops belonging to independent repair banners, tier-2 approved ones included. This is where the "25,000" comes from, the figure that circulates as though it covered every repair shop in the country. | 2024-2025 | By banner, across member shops only. An independent outside any network is never counted. |
| Aggregated content, "how many repair shops" pages | 104,000 | A recompilation of ranges lifted from other pages. The total and its subsets do not add up, and no year is attached to any figure. | Updated June 2025 | Six categories, all in wide overlapping ranges, with no method and no dated source. |
| Scalon, this survey | 101,989 | Locations read one by one, across two independent sweeps of the territory, cross-checked and then settled by hand on the edge cases. Deduplicated on unique phone numbers. | Surveyed July 2026 | Seven segments, three scope tiers, 96 districts, listings and unique phone numbers kept apart. The reason for the sort is written on every row. |

The column that matters is not the one with the figures, it is the last one. A number without a breakdown cannot be worked with: knowing there are 128,000 companies in the branch does not say how many are shops, nor which of them repair light vehicles, nor which are reachable. That is the whole point of this census, and the full breakdown is in section 04 of the report.

## 02 · What is the activity code for a repair shop?

There is not one. In France the activity code is assigned by the national statistics institute at registration, based on the declared activity. For automotive aftersales, the classification hands out at least five of them, and they do not overlap.

- **4520A** — Maintenance and repair of light motor vehicles. The repair shop in the everyday sense: mechanical, body, tires and electrical work on cars and light commercial vehicles.
- **4520B** — Maintenance and repair of other motor vehicles. Heavy trucks.
- **4511Z** — Sale of cars and light motor vehicles. The code for dealerships and franchised agents, whose workshop appears nowhere.
- **4532Z** — Retail sale of automotive equipment. Auto centers, and some tire shops.
- **7120A** — Vehicle inspection.

None of those codes covers the trade, and no combination of them does either. Targeting built on 4520A alone leaves out shops registered under retail or equipment. A union of all five drags in head offices, chains with centralized purchasing and closed businesses. Either way, the code answers an administrative question, not yours.

So this census does not start from a classification filter. It starts from the territory, and reads locations one by one.

## 03 · Activity code and business code: what is the difference?

In practice, none. It is the same code, four digits and a letter. In France two names circulate for it: one designates the principal activity carried out, meaning what the code qualifies, the other designates the national classification of activities, meaning the catalog the code is drawn from. The statistics institute assigns a code by taking it from that catalog. The two spellings cover no substantive difference, and a repair shop carries the same 4520A whichever one is written.

One nuance does matter for targeting: the code of the company and the code of the individual location can differ. A group registered under retail may operate a maintenance shop. This census counts locations, not companies.

## 04 · The three tiers

Each tier contains the previous one. The two right-hand columns do not carry the same number because one location can appear under several online listings: the unique phone number is the honest measure of a prospecting file.

| Tier | What it contains | Listings | Unique phones |
| --- | --- | --- | --- |
| Core of the trade | General mechanical and body work | 60,059 | 56,300 |
| Extended shops | + tires, glass, diagnostics, specialists | 75,800 | 70,000 |
| Full market | + franchised agents, inspection, roadside assistance | 101,989 | 89,300 |

Going from the core of the trade to the full market is **70% more listings and 59% more phone numbers**. That is the concrete measure of what a properly defined scope is worth, and the order of magnitude a sales team loses when it sticks to the most obvious label.

### The breakdown by type

The three tiers are totals. Here is what is inside them, segment by segment. None of the sources in section 01 goes down to this level: it is the one thing a census does, and that an activity code or a branch scope never will.

The "status" column distinguishes three kinds of number, because they are not worth the same. **Surveyed**: the segment was counted in its own right. **Derived**: the number falls out of a subtraction between two surveyed tiers, it is exact but it was not measured directly, and the arithmetic is written out. **To break down**: the segment exists and has no publishable count yet, which we would rather write than paper over.

| Segment | What it contains | Listings | Status |
| --- | --- | --- | --- |
| General mechanical | Maintenance and repair shops, all makes. | — | To break down |
| Body and paint | Body shops, including insurer-approved ones. | — | To break down |
| Core of the trade | The two segments above, counted together. | 60,059 | Surveyed |
| Tires, glass, diagnostics, specialists | Single-service shops, networks included. | 15,741 75,800 − 60,059 | Derived |
| Extended shops | Core of the trade and specialists combined. | 75,800 | Surveyed |
| Franchised agents | Filed as dealerships by automatic categories, although they repair vehicles and buy parts. No export by category returns them. | 13,455 | Surveyed |
| Inspection and roadside assistance | Approved inspection centers and tow operators. | 12,734 101,989 − 75,800 − 13,455 | Derived |
| Full market | Everything above. | 101,989 | Surveyed |

Two segments are still to be broken down, general mechanical and body work, currently counted together under the core of the trade. They will be separated on the next pass: a paint distributor and an engine parts distributor are not looking for the same shops, and delivering both with the same total would do to them exactly what this census criticizes the other sources for.

## 05 · The 13,292 repair shops nobody counts

This is the most counter-intuitive result of the survey. A business of the "Dupont Garage, Renault agent" type is filed as a **dealership** by the automatic categories of online maps, never as a **repair shop**. It repairs cars all day long, and it buys parts.

A naive export by category does not see them. We identified them by cross-referencing, using two independent methods:

- **9,916** explicitly declare repair work in one of their secondary categories.
- **3,539** carry a shop name or sign, of the "Garage…" or "Body shop…" type, without declaring it.

After merging duplicates between the two methods, **13,455 listings** come out, of which **13,292 with a phone number**. That is thirteen percent of the full market, invisible to anyone sorting by category. And these are not the small ones: they are often the best equipped and the biggest buyers in their area.

What it costs not to see them

A parts distributor that builds its list on the "repair shop" category works on 87% of its market while believing it covers all of it. The error never shows, since nothing in the file flags the absence.

## 06 · Breakdown by district

The survey covers the 96 districts of mainland France. Density per head of population says more than raw volume: it shows that rural areas are not under-served, they are simply less concentrated.

| District | Core of the trade | Full market | Per 10,000 people | Network share |
| --- | --- | --- | --- | --- |
| Nord (59) | 2,418 | 4,106 | 15.6 | 31% |
| Bouches-du-Rhône (13) | 1,874 | 3,189 | 15.4 | 29% |
| Rhône (69) | 1,611 | 2,740 | 14.1 | 36% |
| Gironde (33) | 1,305 | 2,218 | 13.3 | 33% |
| Loire-Atlantique (44) | 1,118 | 1,901 | 12.9 | 35% |
| Paris (75) | 1,042 | 1,673 | 7.6 | 44% |
| Creuse (23) | 128 | 218 | 19.1 | 11% |

A seven-row extract. The full table covers all 96 districts, each with its own page and the detail of its ten largest towns.

## 07 · What we deliberately ruled out

A census whose exclusions are unknown cannot be verified. So we publish ours. **6,512 listings** were examined and then taken out of scope: parts trading with no workshop, car washes, garden machinery, motorcycles, industrial machining.

That is around 6.5% of the full market, and no single pocket exceeds 1.4% on its own. That is what lets us state that no whole category was overlooked by accident: if a trade were missing, it would show up as a hole, and holes are visible.

## 08 · The method

Two independent sweeps, run in parallel and compared at the end. The first is systematic and map-based: the territory is divided and covered cell by cell, until saturation, meaning until one more cell returns no new business. The second works by keywords, town by town, using the real vocabulary of the trade rather than that of the classifications.

The two sets are then cross-checked against each other, and against the official registry, district by district. Where the two methods diverge, a person reads the listing, the website and the reviews, and rules with a written reason. Saturation was reached in early July 2026.

The registry lists around **107,000** active businesses in light vehicle repair, against around 96,200 listings open online in the relevant categories. The gap of roughly ten thousand is explained by sole traders with no commercial premises: legally in existence, commercially unreachable, and therefore out of scope for a team prospecting by phone or in the field.

## 09 · The limitations, stated

No census is exact down to the last business, and claiming otherwise is the surest way not to be believed.

- **4 to 8% of closures** are not yet recorded online at the time of the survey. That is the annual attrition rate of the scope, and the reason an unmaintained list degrades fast.
- **Around 3% of the full market** is outside light vehicles: marine, agricultural, motorcycles. Those listings are labeled as such, therefore filterable, and not deleted.
- **Listing duplicates** are detected and merged on delivery. That is the gap between the listings column and the unique phone numbers column.
- **The scope is mainland France.** The overseas territories are surveyed separately.

One check reassured us about the reliability of the closures flagged: out of **3,524 listings** marked closed, fewer than 6% carried a review less than six months old. When the listing says closed, it is almost always true. The reverse is not guaranteed.

## 10 · How to target repair shops without going through the activity code

This survey says how many there are. It does not say which ones you care about. In that same market, a parts distributor that rules out chains with centralized purchasing, heavy trucks, agricultural machinery and closed businesses comes down to a scope of roughly **27,000 genuinely addressable accounts**, contacts included.

That is the difference between a market figure, which is public and will stay public, and a qualified list, which is built on your brief and has value only for you.

Get the full survey

Eleven sections, including the district-level reading, the method and the limitations.

Full survey · PDF

## The survey runs to 8 pages, we will send it to you

The three figures are in the title of this page, and you can quote them without asking for anything. What comes after, how we went from 25,000 to 101,989, what we counted, what we ruled out, what stays uncertain, fits in one document. It goes out by email, right away.

1. 01–03Why published figures run from 25,000 to 108,000The word "repair shop" covers five trades. The five activity codes involved, and why none of them covers the market.
2. 04The three tiersListings and unique phone numbers, tier by tier. From the core of the trade to the full market, that is 70% more listings.
3. 05The 13,292 repair shops nobody countsThe two cross-referencing methods, and what it costs to work 87% of a market believing you cover it.
4. 06–07The district reading, and the exclusionsDensity per 10,000 people, network share, and the 6,512 listings taken out of scope, pocket by pocket.
5. 08–09The method and the limitationsThe two independent sweeps, the saturation criterion, the rate of unrecorded closures. Written out in the open.
6. 10–11Targeting without the activity codeHow you go from 101,989 locations to roughly 27,000 addressable accounts, and how to cite the survey.

Free, no conditions. The counts stay published under a CC BY 4.0 license: you can reuse the figures from the survey, including commercially, citing Scalon and the survey date.

## Citing this census

Reuse these figures freely, including in commercial work. Nothing to ask for, no agreement to sign. We ask for two things: **the name Scalon** and **the survey date**. The date matters as much as the name, because this census is redone every year and a count with no date cannot be verified.

> According to the [Scalon census of July 2026](https://scalon.fr/en/observatory/car-repair-shops), mainland France has 101,989 vehicle maintenance and repair businesses, of which 60,059 at the core of the trade.

If you aggregate, filter or recalculate, say so: *"author's calculations based on Scalon, July 2026 survey"*. What is not allowed is presenting a reworked figure as ours.

Published under a [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/) license, which covers the counts published here and in the report, and nothing else: not the lists of businesses, not the named data, not the qualification columns delivered to clients. If you have a doubt about a particular case, [write to us](mailto:study@scalon.fr).

## Do you sell into this market?

We build a sample of 100 qualified businesses in your scope, free, so you can judge for yourself.

[Book a call](/en#call) [Size my addressable market](/en/tools/local-tam)

Coming next in the Observatory

[All the censuses →](/en/observatory) [The auto parts case study →](/en/cases/auto-parts)