HUD fair market rent data, explained

What fair market rent means

Every year HUD publishes a Fair Market Rent (FMR) for each metro area and many counties. The FMR is an estimate of the 40th percentile gross rent, which means shelter rent plus utilities, for standard-quality units in a local market. In plain terms: 40% of typical units rent for less than the FMR, 60% rent for more. HUD refreshes the numbers each fall for the coming fiscal year.

The number matters because housing agencies use it to set payment standards for the Housing Choice Voucher program, the ceiling on what a voucher covers in a given area. Landlords, analysts, and property teams use it as an official rent benchmark that applies the same way everywhere.

One FY2026 change matters for anyone searching for fair market rent by zip code: many metros now use Small Area FMRs (SAFMRs), which HUD sets at the ZIP-code level instead of one number for the whole metro. A 2-bedroom in one ZIP can carry a different FMR than the ZIP next door. The ZIP-level file is where those differences live.

What is inside the HUD FMR dataset

The FY2026 pack on the HUD FMR dataset page has three CSVs, all amounts in monthly dollars:

• fmr-by-zip-2026.csv, 51,895 rows. One row per ZIP code, built from HUD's FY2026 Small Area FMR file. ZIPs in the same metro share the metro FMR unless HUD assigned that ZIP a SAFMR.
• fmr-by-county-2026.csv, 3,229 rows. County rollup with median and max FMR per bedroom size. Where a county maps to more than one HUD area, the area_rows column shows how many source rows were aggregated.
• state-summary.csv, 52 rows. Per state or territory: ZIP count, 2BR median, the lowest and highest 2BR FMR among its ZIPs, and the highest- and lowest-rent county.

Every file covers five bedroom sizes, 0BR through 4BR. Dollar amounts are integers, rounded to the nearest dollar exactly as HUD published them.

Every column in fmr-by-zip-2026.csv

10 columns, one row per ZIP code.

Two quirks worth knowing before you query. First, HUD's published payment standards are 90% and 110% of the FMR in the same row, so you get them with one multiplication. Second, some areas, notably parts of Texas, fall back to state or national minimums in HUD's methodology. The published numbers already reflect that, so no adjustment is needed on your side. Notes like these are what a data dictionary is for; what a data dictionary is shows excerpts from this pack's dictionary.

Where does the FMR sit against real asking rents? FMR vs actual rent breaks down the percentile, basis, timing, and geography gaps with rows from the free samples.

How to query fair market rent by ZIP code without downloading anything

Downloading the CSV works fine in a spreadsheet. Scripts and agents have a faster path: the dataset-mcp server reads the full 51,895-row file and filters it in memory. Setup instructions are on the For agents page. Weighing the two approaches in general? CSV vs API for data breaks down when each one fits. For a single-ZIP lookup walkthrough, HUD FMR by ZIP code runs one end to end. Working at the metro level instead? HUD metro area list ranks all 649 FY2026 metro FMR areas by ZIP count and shows the 2BR spread inside each one. For a single state run end to end, HUD FMR for Texas counts the 3,247 Texas rows and tables the metro spreads, and HUD FMR for Florida does the same for 1,555 rows with the state-vs-national median gap spelled out.

A real call, filtering the ZIP file for Texas:

query_dataset(
  "hud-fmr-2026",
  {
    "where": [{ "column": "state", "op": "=", "value": "TX" }],
    "columns": ["zip", "area_name", "fmr_2br"],
    "limit": 5
  }
)

Output from a September 2026 test run of server v1.1.0, matching 3,247 Texas ZIP rows:

The cold call, including the one-time file download, took 367ms. The next page came back in 265ms from cache. Each call returns up to 100 rows with a next_offset for paging, and the server verifies the file's SHA-256 on every load. A companion get_stats tool returns row counts, min and max, and top values for any column.

Get the data

Start with the free 22-row sample to check the schema, then take the full pack when it fits. The full pack carries all three CSVs (51,895 + 3,229 + 52 rows), the data dictionary, and source checksums.

Checkout and download run through Getly, part of the pipeline documented in how we sell public datasets. Source: US Department of Housing and Urban Development, FY2026 Fair Market Rent release, huduser.gov. US government data, public domain.