Purine database
The purine dataset, as a file
830 measurements of total purines in milligrams per 100 g, grouped into 372 foods, with the publication printed on every row. Two files, no account, nothing to sign up for. If you are building something that needs purine figures, this is the same data the 372 pages on this site are rendered from.
830 rows · UTF-8 · the JSON carries the field and source definitions inline.
What the columns mean
| Field | Type | Meaning |
|---|---|---|
| name_en | string | English name of the food as measured, including the preparation. |
| category | string | One of seafood, meat & poultry, vegetables, soy & legumes, grains & staples, fruit, dairy & eggs, nuts & seeds, drinks, alcohol, other foods. |
| purine_mg_per_100g | number | Total purines in milligrams per 100 g of the food as described. Not per portion. |
| band | string | The six-band label used on this site. Derived, not measured — see the classification page. |
| source_id | string | Which publication the measurement comes from: NHC-2024 or USDA-ODS-NIH-2.0. |
| source_citation | string | Full citation for that publication, repeated on every row so a copy cannot lose it. |
| page_url | string | The page on this site that covers the food. Empty where no page is published for it yet. |
The band column is the one piece of judgement in the file: six levels — low (under 30 mg), fairly low (30–75 mg), fairly high (75–150 mg), high (150–200 mg), very high (200–300 mg), extremely high (over 300 mg) — each cut line drawn by a published guideline, but the combination is ours. How the bands are drawn. Drop the column if you would rather band the values yourself; the measurements do not depend on it.
Where the measurements come from
- NHC-2024
- Dietary Guideline for Adults with Hyperuricaemia and Gout (2024 Edition), National Health Commission of the People's Republic of China, appendix 1.398 measurements · publisher
- USDA-ODS-NIH-2.0
- USDA and ODS-NIH Database for the Purine Content of Foods, Release 2.0 (2025), Agricultural Research Service, U.S. Department of Agriculture.432 measurements · publisher
Nothing is estimated, averaged across sources, or matched between them by name similarity. Where both publications measured something that looks like the same food, both rows are kept with their own citation, because they measured different preparations of it and the figures are not interchangeable.
How to cite it
If the compilation is useful, this is the line we would like back:
Breeze, "Purine content of foods" dataset, 830 measurements, https://breezegout.com/purine-foods/dataset. Measurements are reproduced from the publications named in each row and should be cited to them.
The measurements themselves should be credited to the publication in the row, not to us. We compiled, normalised to one unit and banded them; we did not measure anything.
Terms
The measurements belong to the two publications above and carry whatever terms those publishers set: the USDA and ODS-NIH release is a work of the U.S. federal government and is in the public domain in the United States, and the National Health Commission guideline is a Chinese government publication whose terms are set by its publisher, not by us. Anyone republishing either set of figures should go to the original and check.
The compilation on top of them — the grouping into search terms, the single unit, the six-band scale and these files — is free to use, including commercially, with attribution to Breeze. One condition, and it is the reason this file exists: keep the source column. An uncited purine chart is exactly what this site was built to replace, and a copy of this data without its citations becomes one the moment it leaves here.
This is a summary. The full terms, including the source attribution for both publications, live on one page and one page only: Using the data. That is the URL the machine-readable license and the JSON file both point at.
What this dataset cannot tell you
It reports total purines per 100 g of food, which is not the same question as what will happen to your uric acid. Most urate is produced internally rather than eaten, and the effect of diet alone is modest next to urate-lowering medication. Beer is the standing example: under 15 mg per 100 ml, and still associated with flares, because alcohol also slows how the kidneys clear urate.
Three more limits worth knowing before you build on it. Per 100 g is not per portion — a 30 g garnish of a 400 mg food contributes less than a 250 g plate of a 100 mg one. Total purines are not split into the four bases, and the bases are not equivalent in their effect. And Chinese names from the NHC appendix are not included, so matching a row back to the original by name means going through the English name we use.
Reference information only. Purine values describe foods, not people: they cannot tell you what is safe for you to eat, and nothing here diagnoses or treats gout. Talk to your doctor or dietitian about your own diet and medication.
Questions about the data
What exactly is in the file?
830 measurements of total purines in milligrams per 100 g of food, each with the publication it came from. 398 measurements are from appendix 1 of China's National Health Commission 2024 dietary guideline for hyperuricaemia and gout, and 432 are from the USDA and ODS-NIH Database for the Purine Content of Foods, Release 2.0 (2025). The preparation is part of the name, because it changes the number: dried anchovy and fresh anchovy are separate rows, not one averaged 'anchovy'.
Can I use it in my own app or article?
Yes. The measurements are not ours to license — they belong to the two publications named on every row, and anyone republishing them should check the original publisher's terms. What we ask for is the citation: keep the source column, and credit this compilation if it saved you the work of building one. That is the whole point of shipping the sources inside the file rather than beside it.
Why are some numbers different from other purine charts?
Three reasons, and none of them is that someone is wrong. Charts differ on preparation (fresh versus dried versus grilled), on whether they report total purines or only some of the four purine bases, and on national thresholds for what counts as 'high' — the Chinese guideline draws its top band at 150 mg per 100 g while the Japanese one does not call a food high until 200. This file reports total purines per 100 g and names the preparation, so the comparison is at least like for like.
How often does it change?
Rarely, and only when a source publishes a correction or a new release. The JSON carries a version field set from the commit date of the underlying data, so a rebuild that changed nothing does not claim to be a new version. There is no API and no versioned archive — if you need the data to be stable, keep the copy you downloaded.
Would rather just look something up?
The same data is browsable as 372 food pages, each listing every measurement behind it, or you can compare two foods side by side.