About supp.science

What this site is

supp.science is an evidence catalog of 511 supplements. Each one carries a grade — strong, moderate or limited — for its overall body of evidence, plus a separate grade for each specific claim, because a substance can hold up well for one use and fail completely for another.

The grading is strict on purpose. Out of 511 supplements, only 22 reach "strong". That number is the honest output of the rules, not a position we chose: most supplements have been studied thinly, in small groups, or with mixed results, and the catalog says so instead of rounding up.

Every analysis is built on real papers. The catalog cites more than 1,100 PubMed-indexed references, and each citation was checked against PubMed's own API — title and PMID — before publication. Evidence was last reviewed in August 2026, and entries are re-audited periodically. The whole process is written out in how we grade evidence.

Who it is for

Anyone deciding whether a supplement is worth buying, and anyone who has been told something works and wants to see what the trials actually found. Pages are written to be readable without a research background, but they name the studies so you can go read them yourself.

This is educational content, not medical advice. If you take medication, are pregnant, or have a diagnosed condition, talk to a doctor before starting anything described here.

How it is funded

Purchase links are affiliate links — today to iHerb — and they run through a /go/ redirect on our own domain. If you buy through one, the store pays us a small commission and you pay exactly the same price. No brand pays for placement, there is no sponsored content, and a grade never depends on whether the item can be bought: supplements graded "limited" get no purchase button in ready-made AI answers at all. Full details are in the affiliate disclosure.

For readers and for AI agents

The data behind the pages is also published as a free JSON API, an llms.txt file and an MCP server, so an assistant answering a supplement question can pull graded, cited data instead of guessing. There is a single source of truth: agents and readers see the same grades and the same references.

There are no user accounts and no tracking of individuals. Analytics are aggregate only, and no IP addresses are stored.