Methodology

Nobody agrees how many breweries there are, including us

9,344, 12,238, or 7,204, depending on who is counting and what they think a brewery is. Here is every number we hold and exactly what each one means.

The Brewers Association counts 9,344 US breweries operating in June 2026. Our graph holds 12,238 brewery records worldwide. Open Brewery DB, the open dataset most apps are built on, lists roughly 11,700.

None of those is wrong. They are answers to three different questions, and almost nobody says which question they asked.

Every number we hold, and what it means

Why we are short of the BA, and why we say so

Our US layer is built mostly on Open Brewery DB, which is community-maintained and lags new openings. Of our active US records, the overwhelming majority carry an OBDB identifier and only a few hundred carry a TTB brewer permit number.

The obvious fix is to reconcile against the federal brewer permit list. That turns out to be harder than it sounds. The TTB permit data we hold is dominated by importer and wholesaler permits — tens of thousands of them, state-prefixed — against a few hundred brewer-prefixed permits. There is no tidy public file of every operating US brewery with a permit number and an address. If there were, everyone's count would agree.

So we are undercounting active US breweries by something like 20%, and I would rather write that down than round it away.

The definitional problem underneath

Before you count breweries you have to decide what one is, and every sensible answer produces a different total.

A brand that has beer made under contract at someone else's plant: one brewery, or zero? A brewpub group with six locations and one brewhouse: six, or one? An alternating proprietorship where two companies take turns on the same kettle: two? A production facility with no taproom is a brewery to the BA and to the TTB, and is nothing at all to a person planning a Saturday.

We answer that last one by keeping two numbers and labelling both. The active count is the industry answer. The visitable count is the drinker's answer. Collapsing them into one figure would make our marketing easier and our data worse.

“Active” is doing a lot of unearned work

Here is the part I did not expect. We recently checked every active US brewery whose address we could not place on a map, which meant visiting a few dozen brewery websites.

Eighteen of them no longer have a working website. Dead domains, hard 404s, servers that time out. One brewery in the set is listed in our graph in a city that the company's own site no longer includes among its locations.

Every one of those is currently counted as active — by us, and probably by everyone else working from the same open data. A dead website is not proof of a closure. It is a strong hint, and it is the kind of hint that a count of 9,344 quietly contains an unknown number of.

Closures lag in every dataset, because closing is not an event anyone files. Openings get press releases. Closings get a padlock.

Which number should you use

What we publish

The counts on this page are read live from the graph when the page builds, so they move when the data moves. The brewery index and the A–Z directory are the full lists, state pages like California carry regional counts, and the methodology page holds the sourcing rules.

If you find a brewery we have wrong — closed, moved, missing — that correction is worth more to us than another thousand rows.

Method

Counts read from the Hopgrail brewery graph at build time. The graph merges Open Brewery DB snapshots with TTB permit and label records, resolving conflicts by field with recorded provenance. “Visitable” = US, status active, and brewery type not contract, alternating proprietor, planning, bar, or closed. BA figures from the 2026 Midyear Report. Last checked 3 August 2026.

See the underlying records

Every brewery we hold, with its filings and its sources. The map for the visitable ones. And the midyear numbers for what the count is doing this year.