Guide
What is a NACE code? NACE Rev. 2.1 explained
NACE is the EU's industry classification. How the codes are built, what changed in NACE Rev. 2.1 (new section K, relettered sections) and when each country switches.
6 min read · updated
A NACE code says what a business does, in one standard form shared by every EU member state. If you build lead lists from European business registers, it is the most useful filter you have, and since 2025 it has changed in ways that quietly break older filters.
What NACE is
NACE stands for Nomenclature statistique des activités économiques dans la Communauté européenne, the statistical classification of economic activities in the European Community. Eurostat maintains it, and national statistics offices and business registers use it, or a national version of it, to record each company's main activity.
The current version is NACE Rev. 2.1. It replaces NACE Rev. 2, which had been in use since 2008. Every country covered on this site has moved to a 2.1-based national code, or is in the middle of moving.
A NACE code is not a licence or a legal status. It is a label, usually chosen by the company when it registers and sometimes reviewed later by the statistics office. That matters when you rely on it, and the guide to finding a company's NACE code covers the pitfalls.
How a NACE code is built
NACE is a four-level tree. Each level narrows the one above it.
| Level | Format | Example | Meaning |
|---|---|---|---|
| Section | one letter | K |
Telecommunication, computer programming, consulting, computing infrastructure and other information service activities |
| Division | two digits | 62 |
Computer programming, consultancy and related activities |
| Group | three digits | 62.1 |
Computer programming activities |
| Class | four digits | 62.10 |
Computer programming activities |
NACE Rev. 2.1 has 22 sections (A to V), 87 divisions, 287 groups and 651 classes. NACE Rev. 2 had 21 sections (A to U) and 615 classes.
The section letter is not part of the numeric code. You will never see K62.10 in a register; you see 62.10 and know from the division that it sits in section K. The divisions are numbered across the whole classification, so division 62 always belongs to exactly one section.
When you filter, a higher level includes everything beneath it. Searching for division 62 returns every company coded 62.10, 62.20 or 62.90. Searching for section K returns all of divisions 61, 62 and 63.
National versions: the first four digits are NACE
EU member states may add detail below the class level, but they may not change the first four digits. That is why national codes look different yet convert cleanly:
| Country | National scheme | Example | NACE class |
|---|---|---|---|
| France | NAF | 62.10Z |
62.10 |
| Sweden | SNI | 62.100 |
62.10 |
| Finland | TOL | 62100 |
62.10 |
| Czech Republic | CZ-NACE | 6210 |
62.10 |
| Estonia | EMTAK | 62101 |
62.10 |
| Belgium | NACE-BEL | 62.100 |
62.10 |
Take the first four digits, put a dot after the second, and you have the NACE class. The national extra digit or letter adds detail you can usually ignore when comparing countries. The national equivalents guide walks through each scheme and its quirks.
What changed in NACE Rev. 2.1
The 2.1 revision updates the classification for the digital economy and new activities, and it reshuffles the top of the tree.
A new section K for IT and telecom
In NACE Rev. 2, section J ("Information and communication") held publishing, film, broadcasting, telecoms, programming and information services together. In 2.1 it is split in two:
- Section J: publishing, broadcasting, and content production and distribution (divisions 58, 59 and 60).
- Section K: telecommunication, computer programming, consulting, computing infrastructure and other information services (divisions 61, 62 and 63).
Software publishing stays under publishing: class 58.29 "Other software publishing" is in J, while 62.10 "Computer programming activities" is in K. A "software companies" filter usually needs both.
Every section after J moves one letter
Because K is inserted, the old K to U each move down one letter, and a new section V appears at the end:
| NACE Rev. 2 | NACE Rev. 2.1 | Title in 2.1 |
|---|---|---|
| J (part) | J | Publishing, broadcasting, and content production and distribution activities |
| J (part) | K | Telecommunication, computer programming, consulting, computing infrastructure and other information service activities |
| K | L | Financial and insurance activities |
| L | M | Real estate activities |
| M | N | Professional, scientific and technical activities |
| N | O | Administrative and support service activities |
| O | P | Public administration and defence; compulsory social security |
| P | Q | Education |
| Q | R | Human health and social work activities |
| R | S | Arts, sports and recreation |
| S | T | Other service activities |
| T | U | Activities of households as employers and undifferentiated goods- and service-producing activities of households for own use |
| U | V | Activities of extraterritorial organisations and bodies |
Sections A to I keep their letters. The two-digit division numbers mostly keep their meaning across the revision, which is why filters written in division numbers survive better than filters written in letters.
The trap is obvious once you see it: "section M" meant professional and scientific services in Rev. 2, but means real estate in 2.1. A filter of "sections C, J and M" written in the Rev. 2 era becomes C, J, K and N in NACE 2.1. Keep only C, J and M and you lose the IT companies and pick up estate agents instead.
Changes below the section level
Inside the sections, some classes were split, merged or moved. Eurostat publishes an official correspondence table between Rev. 2 and 2.1. Most links are one-to-one: an old class becomes exactly one new class. Others are many-to-many, where one old class spreads over several new ones or several old ones merge.
On this site, legacy Rev. 2 codes are converted only through exact one-to-one correspondences. Where a code has several possible 2.1 targets, the national code is kept as published and the NACE 2.1 field stays empty rather than guessed. That is a deliberate choice: an empty field is visible, a wrong one is not.
When each country switches
National registers do not all switch on the same day. Based on the bulk files we load:
| Country | 2.1-based scheme | Status in the register data |
|---|---|---|
| Finland | TOL 2025 | The register recoded to TOL 2025 around 1 January 2026. Some dormant companies still carry TOL 2008 codes. |
| France | NAF 2025 | Sirene switches on 1 January 2027. Throughout 2026 the stock files already carry a NAF 2025 column next to the old code. |
| Sweden | SNI 2025 | The SCB bulk file already publishes SNI 2025 codes. |
| Czech Republic | CZ-NACE 2025 | Published in the statistical register, with CZ-NACE 2008 still present for some entities. |
| Estonia | EMTAK 2025 | About 71% of entities have an EMTAK 2025 code; about 25% still have only EMTAK 2008. |
| Belgium | NACE-BEL 2025 | Published next to NACE-BEL 2008 in the KBO/BCE open data. |
The practical consequence is that, for a while, a single register can contain both old and new codes. If you load raw register files yourself, check which scheme each row uses before you compare it with anything. Here every company is normalised to NACE 2.1, and the original national code and scheme are stored next to it, so you can always see what the register actually said.
Why this matters for lead lists
If you prospect by industry, the classification is your targeting. Three practical rules follow from the 2.1 changes:
- Use 2.1 codes everywhere. Mixing Rev. 2 letters with 2.1 data silently returns the wrong industries. Look codes up rather than typing them from memory; the NACE code directory lists every 2.1 section, division and class.
- Prefer divisions and classes over section letters. "IT services" is better expressed as divisions 62 and 63 than as "section K", which also includes telecom operators.
- Expect gaps in older data. Companies that have not updated their registration, or whose old code has no exact 2.1 equivalent, may have no 2.1 code. Very broad searches should allow for that.
When you work through Claude, the lookup_industry tool turns plain words such as "software", "logiciel" or "ohjelmisto" into the right 2.1 codes, so you rarely have to handle letters at all. A request like "computer programming companies in Sweden" ends up as a filter on division 62 in Sweden; "management consultancies in France" as division 70 in France. The lead list guide shows the full workflow, and you can browse companies by country to see how industries are distributed in each register.