Data access

Every page of this site is open. The JSON record for each additive, with a JSON Schema describing its shape, and the CSV files below, which are the whole set as far as the licences allow, are given on request.

How to get a key

  1. Send a request: your name, your email address, and what the data is for.
  2. A person reads it and approves or declines it. You are told by email at the address you gave.
  3. If you are approved, you are sent a key. It is valid for a year unless the email says otherwise and can be withdrawn if it is used against the licences of the sources behind the rows.

With a key, send it as a header (preferred) or as ?key=:

curl -H "Authorization: Bearer YOUR_KEY" \
  https://foodadditiveindex.com/api/v1/additives/citric-acid.json

curl -OJ -H "Authorization: Bearer YOUR_KEY" \
  https://foodadditiveindex.com/data/additives.csv

A key allows 20,000 requests a day. Without one, these paths answer 401 and say where to ask. What you write in the request, and what is kept, is set out in theprivacy statement.

The files

FileRowsWhat it holds
additives.csv1,189One row per additive: slug, primary name, E number, INS number, CAS number, InChIKey, PubChem CID, functional class, starting material, and the URL of its page.
permissions-eu.csv18,137One row per provision: the additive, the jurisdiction, the status, the food category in its own system, the maximum level and unit, the conditions as written, the citation reference and URL, the source key and the date it was retrieved. EU.
permissions-gb.csv18,003The same columns, for GB.
permissions-anz.csv14,486The same columns, for ANZ.
permissions-ca.csv2,425The same columns, for CA.
permissions-us.csv577The same columns, for US.
permissions-kr.csv462The same columns, for KR.
permissions-jp.csv307The same columns, for JP.

All are UTF-8, comma-separated, RFC 4180 quoted — conditions text contains commas, quotation marks and line breaks, so a naive split on commas will misread them. They are rebuilt with the site, so their contents match the pages exactly. The columns are identical in every permissions file, so concatenating them (minus the repeated header) gives the whole set: 54,397 rows.

One file per jurisdiction rather than one file for everything, because everything is 22 MB and the ceiling for a file served this way is 25 MB. A single file would have worked until the next amendment to the EU's Annex II and then stopped deploying.

What is not in them, and why

11,102 permissions and every reference value and specification are on the site but not in these files. The reason is licensing, not quality: each of these sources permits this site to use its material with acknowledgement, and none of them permits us to hand out a restructured copy of it.

EFSA OpenFoodTox (2026-04-30)
CC BY-ND 4.0 — NoDerivatives forbids sharing a restructured extract.
Codex General Standard for Food Additives (CXS 192-1995)
FAO terms grant non-commercial use with acknowledgement but reserve adaptation rights; a restructured bulk file needs FAO’s written permission.
Combined Compendium of Food Additive Specifications (JECFA), online edition
FAO terms, as for the Codex standard above; acknowledgement covers use, not redistribution as adapted data.
Evaluations of the Joint FAO/WHO Expert Committee on Food Additives (JECFA)
WHO terms of use reserve reproduction of substantial portions for prior written authorisation.

Nothing is hidden by this. Codex provisions, JECFA and EFSA intakes and both sets of specifications are all on the additive pages, and in the per-additive JSON for a key-holder, each quoted with its citation. If you need them in bulk, ask the body that publishes them — FAO at copyright@fao.org, WHO through its permissions page, EFSA through Zenodo, where OpenFoodTox is published in full.

This site's own writing — the summaries, the derivation notes, the interpretations — is also left out. It is not licensed data from anywhere, and publishing it as a dataset would mean setting terms for it that nobody has set.

Terms

Every row carries its source. Reuse is governed by that source's licence, and the credit each one requires is below. Where a row is public domain, no credit is owed; it is given anyway.

anz — CC BY 4.0
Based on content from the Federal Register of Legislation
ca — Open Government Licence – Canada
Contains information licensed under the Open Government Licence – Canada
ecfr — Public domain (US federal government)
eCFR, Office of the Federal Register
eu — CC BY 4.0
© European Union, European Commission food additives database
gb — Open Government Licence v3.0
Contains public sector information licensed under the Open Government Licence v3.0
jp — Public Data License 1.0
e-Gov Law Search (Digital Agency); Japanese Law Translation (Ministry of Justice); MHLW. Edited by this site; not an official publication
kr — Not protected by copyright (Copyright Act of Korea, Art. 7); free use under the Public Data Act
Ministry of Food and Drug Safety notification, via the National Law Information Center (law.go.kr). Edited by this site; not an official publication; the Official Gazette prevails

This site is non-commercial, permanently, and that is a condition of using some of its sources — see Sources. That condition binds this site; what binds you is the licence on the rows you take, which for everything in these files permits commercial reuse except where a government licence says otherwise.

The files are a convenience, not an authority. A provision is what the instrument says, and the citation on every row is there so you can check it. Where this site has matched a substance to a page wrongly, the row inherits that error: the unmatched names page records what did not match at all, and Method explains how matching works.