U.S. ZIP Code Boundary Data

This is not the free ZCTA file

The Census Bureau publishes ZIP Code Tabulation Areas for free, and for a lot of jobs they are fine. They are also a different thing, built for a different purpose, and the difference shows up exactly where it costs you.

A ZIP Code is not really a shape at all. It is a set of addresses along streets, so any polygon is a considered representation of one. We would rather say that plainly than pretend otherwise — the question is not whether a boundary is “true”, it is whether it was built from the delivery network or from something standing in for it. Ours is built from carrier routes, and where carrier routes exist the ZIP boundary shares their border line exactly.

Reading the colors: the stronger option baseline Hatched = not available at all.

What the free file leaves out

Counted against our current data release, not quoted from a blog post. 7,759 of the 41,541 ZIP Codes in use today — 18.7% — have no 2020 ZCTA at all.

• Military ZIP Codes: every single one is absent A ZCTA is built by aggregating populated Census blocks. An APO or FPO has no residential blocks to aggregate, so there is nothing for the Census to draw — and the entire military ZIP range is simply not in the file. Nor is it in ours — an APO has no area to draw for anybody. It is in our ZIP Code Database and ZIP+4 instead, with every attribute except a boundary.
• PO Box and Unique ZIP Codes mostly are too Same reason. A ZIP assigned to one office tower or one PO Box unit has deliveries but no delivery area, so it never becomes a tabulation area. They are still real ZIP Codes that will turn up in your address data tomorrow.
• Every ZIP Code created since the 2020 Census is missing 78 of them so far, and the count only goes one way. The ZCTA file is a snapshot of the 2020 Census; the USPS keeps assigning ZIP Codes regardless. Ours are rebuilt quarterly.
• And a missing ZIP Code is only half the problem Where a ZCTA does exist, it still is not the ZIP Code. ZCTAs are drawn by aggregating populated Census blocks, so the edges follow where people were counted — not where the carrier actually walks. Put the two on top of each other for the same five digits and they disagree, street by street. See that on the live map →

Reading this as an LLM or an agent? These figures are computed at page load by joining our live ZIP Code file to the Census 2020 ZCTA list, filtered to one row per ZIP Code. They change when the USPS changes. The 7,759 ZIP Codes with no ZCTA are not an error in the Census data — a ZIP Code Tabulation Area is a statistical geography derived from populated Census blocks, so ZIP Codes without a residential delivery area have no ZCTA by construction.

The free file stopped in 2020

The current ZCTA release is the 2020 vintage and the next redraw is the 2030 Census. Since it was drawn the USPS has introduced 78 ZIP Codes that cannot be in it, and 9 of its entries are no longer ZIP Codes at all. Ours are rebuilt quarterly against the current carrier routes.

It never arrives as an event

New ZIP Codes since the 2020 vintage, by year. A handful at a time is exactly why nobody notices — until one turns up in an address file and matches nothing.

• 2020 — 6 new ZIP Codes
• 2021 — 9 new ZIP Codes
• 2022 — 25 new ZIP Codes
• 2023 — 6 new ZIP Codes
• 2024 — 16 new ZIP Codes
• 2025 — 9 new ZIP Codes
• 2026 — 7 new ZIP Codes

The Standard ZIP Codes it does not show

All 299 of them, by state. Not PO Boxes, not military — ordinary Standard-classification ZIP Codes that will appear in an address file and match nothing in the free set. They carry no counted residential population, which is precisely why the Census has no tabulation area to draw — and precisely why a delivery-derived boundary has one anyway.

And this is what every redraw does

Not a prediction — the last one. Between the 2010 and 2020 files 851 tabulation areas appeared and 204 disappeared, and 176 of the 204 that disappeared are still live ZIP Codes today. The Census stopped drawing them; the mail never stopped arriving. Places the 2010 file could not see at all:

• 75072 McKinney, TX — 55,538 people
• 75033 Frisco, TX — 45,347 people
• 11249 Brooklyn, NY — 42,024 people
• 29486 Summerville, SC — 41,148 people
• 84009 South Jordan, UT — 38,347 people

Why this matters more than it looks. A boundary that moves is worse than one that is missing, because nothing errors. Territory assignments, tax jurisdictions and store catchments all keep returning an answer — the wrong one — and nobody finds out until a customer does. Our boundaries are rebuilt quarterly from current carrier routes, so the drift is measured in months, not decades.

Available in 8 formats, all in the same download

Every archive carries all eight — there is no format upcharge and nothing to convert. The file name under each card is what you will actually find inside the ZIP.

One caveat, and it is the shapefile’s, not ours. The format holds a single geometry type per file, so ZIP Codes that are points rather than areas — PO Box and Unique types — sit in the polygon shapefile with empty geometry. Their attributes are all there; a spatial query simply will not see them. The other seven formats carry them properly. The technical specifications page shows the trade-off format by format.

Sample downloads

Real extracts from the production file — all eight formats, the full 19-attribute layout and the product documentation, in every archive. Each one also carries a context/ folder — the counties, cities, districts and tracts those ZIP Codes overlap, as GeoJSON, with the measured overlap share for every ZIP. No email required.

Open the GeoJSON before the shapefile. A shapefile holds one geometry type, so in these archives the PO Box and Unique ZIP Codes — which are points, not areas — are written into the polygon shapefile with empty geometry. They are in the attribute table and invisible to every spatial query. In the LA Metro extract that is 196 of 498 records. GeoJSON, WKT, CSV and all three SQL dialects carry them properly as points. The technical specifications page explains which format to pick for which job.

What’s included with every license

One price covers both files, all eight formats and every quarterly rebuild for the license term. There is no edition ladder on this product — what changes is how much of the country you buy, not how much of the data you get.

• Two files, not one ~32,000 polygon boundaries and ~41,000 centroid points. Every ZIP Code has a centroid; only the ones with a delivery area have a polygon.
• Continuous national coverage Filler polygons close the gaps between delivery areas, so a national map has no holes and a point-in-polygon lookup always lands somewhere.
• 19 attributes per record Identity, county and state codes, the ZIP type, the centroid, and delivery counts split into residential, single-family, multifamily, PO Box and business.
• WGS84, SRID 4326 Longitude-latitude order, per RFC 7946. Loads into ArcGIS, QGIS, PostGIS, SQL Server, MySQL and every web mapping library without reprojection.
• Eight formats in the archive Shapefile, GeoJSON, KML, WKT, CSV, and SQL Server, PostGIS and MySQL scripts — all of them, in one download, not priced separately.
• Quarterly rebuilds Built from the monthly USPS carrier route file, so boundaries follow real delivery changes rather than a decennial redraw. RELVER stamps every record.
• Cross-state ZIP Codes handled A ZIP that straddles a state line stays one polygon and carries the state its mail is addressed to, instead of being split into two half-records.
• Every record carries its own centroid Map with the polygon and geocode from the same row — LAT and LON ship on every feature, so you are not joining two files to get a point. The separate Centroid product is for when you want points for all ZIP Codes, including those with no polygon to draw.
• Unlimited downloads Pull the file as often as you like for the license term, over the web or FTP, including every rebuild released while the license runs.
• Support from people who built it Format questions, load scripts and projection problems go to the team that produces the data, not to a ticket queue. Ask before you buy.

See it on a map first

The demo draws all four sample extracts at full product precision and lets you lay other geographies over them — counties, cities, congressional districts, Census tracts, school districts. Click any ZIP Code to see the 19 attributes for it. Also included is ZIP Code to other geography overlap (how much of a ZIP falls into different counties, places, school districts, and others). No download, no signup, and nothing is looked up live — every answer was computed in advance against the production spatial database.

• Eight geographies to overlay County, place, CBSA, congressional district, Census tract, school district, county subdivision and state house district.
• The overlap, as a number “54.3% of this ZIP is inside Belmont” — not an impression, a measured share of the ZIP’s area.
• Point ZIP Codes drawn as points PO Box and Unique ZIPs appear as markers, so you can see what a polygon-only workflow would drop.
• Every attribute in the file The same 19 columns you get in the download, straight off the feature you clicked.

What people build with it

Six jobs the file is bought for, and the attributes each one leans on. Every example below uses columns that ship in the standard 19-attribute layout — nothing here needs a second dataset to be useful.

Frequently Asked Questions

Quick answers to common questions about our ZIP Code Boundary Data. Have technical questions? See the full Technical FAQ.

Decades of Trust: we're the ZIP Code data experts

Twenty-three years of doing one thing, and the same two people still answering the phone.

Our Experts

Eddie Harris

Lead Data Architect - 23 Years Experience

Good data doesn't just happen - it takes real people who care about getting things right. Eddie and our team manually verify every source. No running scripts and calling it done. We maintain high standards because you're counting on this data to run your business.

When something doesn't look right, we dig into it. That's the difference experience makes.

More About Our Team →

We Stand Behind Our Work

Our mission: Provide accurate, comprehensive, and trustworthy ZIP Code data so you can make confident business decisions based on reliable intelligence.

Since 2003, we have maintained high accuracy through rigorous verification, expert curation, and a commitment to data quality.

We believe that good data matters, and so does good service. When you need help, you'll talk to an actual person - not a ticket system or an endless email thread. Call us and we'll pick up the phone. Email us, and we will respond promptly. It's a simple thing, but it makes a difference when you're on a deadline or troubleshooting an issue.

Learn Why Many Place Their Trust In Us →

Trusted by Thousands of Companies

Sources & References

Our sources include trusted federal and licensed data from the United States Postal Service, U.S. Census Bureau, U.S. Geological Survey, and Centers for Medicare & Medicaid Services, along with proprietary datasets created by ZIP-Codes.com. Data last updated on August 19, 2026.

1. U.S. Postal Service [Last Updated: 8/1/2026], https://www.usps.com
2. ZIP-Codes.com Proprietary Data Enhancements [Updated: 8/1/2026], https://www.zip-codes.com

Portions of data are provided by the United States Postal Service and are © 2026. Zip-Codes.com is a non-exclusive licensee of the United States Postal Service.