The Sirene database: SIREN, SIRET and what French public data really contain

What the Sirene database is

The Sirene register, kept by INSEE, the French statistics office, lists businesses, associations and public bodies together with their establishments. Its open version is published on data.gouv.fr as the “Base Sirene des entreprises et de leurs établissements” dataset (in French): licence Licence Ouverte 2.0, monthly frequency, direct download without an account (checked on 4 November 2026).

The dataset comprises six monthly stock files, as ZIP (CSV) and Parquet: legal units, establishments, historical values of legal units and of establishments, succession links and duplicate SIRENs. According to the dataset description, files published from the 1st of the month are a snapshot of the register on the last day of the previous month; ceased units and closed establishments are included. The legal units file of 1 November 2026 is about 985 MB compressed. data.gouv.fr announces that CSV deliveries will stop in the second half of 2027, leaving Parquet only.

The Licence Ouverte 2.0 allows reuse, including commercial reuse, provided you attribute the information. For the Sirene database, the licence gives this example: state the licensor’s URL, www.insee.fr, and the date of the last update of the information reused.

SIREN and SIRET

According to INSEE’s definition of the SIREN number, “the SIREN number is a nine-digit identifier assigned to each legal unit. The first eight digits have no meaning. The ninth digit is a check digit for the validity of the number.” It is assigned only once and never reused.

The SIRET number identifies an establishment: 14 digits, namely the SIREN of the legal unit followed by the NIC (internal classification number), made up of a four-digit sequence number and a check digit. A business therefore has one SIREN but as many SIRETs as it has establishments, current or closed.

A practical tip: store SIREN, SIRET and postcode as text. A spreadsheet that converts them into numbers drops the leading zeros.

What a record contains

For the legal unit: SIREN, date of creation, name (or surname and first names for a sole trader), legal category, main activity (APE code in NAF rev. 2), headcount band, enterprise category, administrative status (active or ceased) and the NIC of the head office. For each establishment: SIRET, address, trading name, activity, status (open or closed) and whether it is the head office.

Since 16 December 2025, the stock files also contain, for information only, the future activity code in NAF 2025, which becomes the reference on 1 January 2027 (see NAF and APE codes). Headcount bands were updated on 1 October 2026 with values as at 31 December 2024.

What the database does not contain: phone, email, website. INSEE does not disseminate data on the legal representatives of legal persons either.

Some figures

We counted the StockUniteLegale file of 1 November 2026 line by line:

• 30,280,224 legal units in total, of which 16,833,577 active;
• among the active units, 14,467,253 with full dissemination (status O) and 2,366,324 with partial dissemination (status P, 14.1%);
• by legal category, among the active units: 7,845,810 sole traders, 4,341,751 commercial companies, 2,539,031 “other legal persons registered in the RCS” (a category that includes civil companies) and 1,486,034 private-law groupings (including associations).

This total is far higher than the 5,178,283 enterprises counted by INSEE for 2023 in the market, non-agricultural and non-financial sectors (INSEE, published on 26 January 2026, Ésane source). The two figures do not measure the same thing: the statistic counts enterprises in the economic sense within a defined scope, while the register records every legal unit, including civil companies, associations and farms. To size a market, start from the statistic; to build a file, start from the register and then filter.

Partial dissemination (status P)

According to the INSEE press release of 21 March 2023 (in French), legal units and establishments that have exercised their right to object under the GDPR are now partially disseminated: “Their economic and legal data are made available to the public: SIREN and SIRET numbers, main activity, legal category, headcount band… However, the identification details of natural persons (surname, first name, pseudonym) and their location details (any address and location details, except the municipality) remain hidden.”

In the files, these hidden fields appear as [ND]. These units have expressed an objection: firmdata.fr excludes them from matching.

The Recherche d’entreprises API

To look up one business at a time, the Recherche d’entreprises API is open without a key. According to its documentation of the Recherche d’entreprises API and its data.gouv.fr page, it accepts at most 7 requests per second per IP address, does not give access to non-disseminated businesses and is meant for searching a business, not for downloading the full database. It adds information from other sources to the Sirene data, for example active intra-EU VAT numbers (source: DGFiP, the tax authority) and executives.

It also publishes the field opposition_diffusion_commerciale: “Objection to the reuse of the legal unit’s data for commercial prospecting under paragraph 2 of Article 21 of European Regulation 2016/679 GDPR (Article R.123-320 of the Commercial Code)”. The value comes from the national business register (RNE) kept by INPI; null means the unit is not in the RNE. This field is not yet included in firmdata.fr exports.

So where do contact details come from?

A business website must display mandatory legal notices (in French), including, according to the official Entreprendre Service Public portal, an email address and a phone number for contacting the business. That is why “contact” and “mentions légales” pages are a major source of business contact details. Open map data are another: they combine a place, a name and often a phone number and website.

How firmdata.fr uses the Sirene database

Our records come from Overture Maps and the businesses’ websites. Matching against the Sirene database is in progress: matched records will receive SIREN, SIRET, NAF code and legal form, together with the date of the Sirene file used. Units with partial dissemination are not matched: their locations may appear in our files as published by Overture Maps, without any flag. How current the data are is shown on the database status page, the method on the sources and methodology page.

Frequently asked questions

Is the Sirene database free?

Yes. The stock files are published every month on data.gouv.fr under Licence Ouverte 2.0 and can be downloaded without registration. Reuse, including commercial reuse, is free provided you state the source and the date of the last update.

Does the Sirene database contain phone numbers and email addresses?

No. It contains no phone number, email address or website. These contact details come from other sources, in particular the businesses’ own websites.

What is the difference between SIREN and SIRET?

The SIREN (9 digits) identifies the legal unit, in other words the business. The SIRET (14 digits) identifies an establishment: it is the SIREN followed by the 5-digit NIC.

What does dissemination status “P” mean?

Partial dissemination: the unit has exercised its right to object. Its numbers, activity, legal category and headcount band remain public; the names of natural persons and the address, except the municipality, are hidden.

As of 4 November 2026. Sources: data.gouv.fr (Sirene database, Licence Ouverte 2.0, Recherche d’entreprises API), INSEE (definitions, press release of 21 March 2023, Ésane statistics), Entreprendre Service Public, retrieved on 4 November 2026; quotations are our translation. Counts by status and category are from our count of the StockUniteLegale file of 1 November 2026 (source: INSEE, Sirene database, www.insee.fr).