Historical ZIP Code Database - point-in-time records going back to October 2004
Point-in-time ZIP Code data

Historical ZIP Code Database

We have published a ZIP Code Database every month since October 2004, and we kept every one of them. That is the whole product — not a reconstruction and not a model, but the actual file customers downloaded that month, exactly as it shipped. The archive starts at publication number one.

257monthly publications
21years covered
94+data fields
Available by arrangement · call 1-800-425-1169

At a glance

Earliest publication
October 2004
Most recent available
March 2026
Field set
Grows by year — check yours
Edition
Business (Deluxe before Oct 2006)
Formats
CSV, Excel, Access, SQL
Updates
None — frozen by design
Delivered as
One archive per month bought

Choose your historical access

Available by arrangement — call us to discuss what you need. Any month from October 2004 to March 2026, or all 257 of them at once.

Single month

Call for Details
1-800-425-1169
Any one month you name
  • Pick any month from October 2004 to March 2026
  • The complete bundle for that month
  • Formats as published in that era
  • Best for one specific date — an audit, a dispute, a cutoff
Call for Details What’s in one month

Licensing & terms

Internal research license

For internal research only — analysis, historical study and statistics inside your own organization. Not for redistribution, products or services for others, production or operational systems, or use with AI. This is the license that applies to a single month and to the complete archive.

Read the historical database terms

Not ready to choose? Below: what a month contains, what changes across 21 years, a live sample, and ten straight answers.
Need the last few months rather than a past one? The current database is the right product — it ships monthly updates for a year and costs less.

What one month actually is

A single month is not an extract or a summary. It is the complete Business bundle for that month — the main table plus the supplemental files — in every format we published it in.

Buying March 2011 gets you the same download a paying subscriber received in March 2011. Buying the complete archive gets you all 257 of them.
FileWhat it is ColumnsIn the bundle
zip-codes-database-DELUXE-BUSINESS.csvThe main table — one row per ZIP / city / alias93Always
census-acs-2020-2024-zcta.csvAmerican Community Survey, 5-year estimates — re-cut yearly543Apr 2022 onward
census-2010-zcta.csv2010 Decennial Census by ZCTA186Oct 2011 onward
census-2020-zcta.csv2020 Decennial Census by ZCTA161Jun 2023 onward
zip-codes-database-PLACE-FIPS.csvPlace FIPS codes per ZIP25Apr 2015 onward
zip-codes-database-MULTI-COUNTY.csvCounty splits for ZIPs crossing county lines6Dec 2007 onward

Column counts and file names are those of the current release. Earlier publications carry fewer files and fewer columns — the archive grew as the product did, which is exactly what makes it useful for tracking change. In particular, the 24 months before October 2006 are the Deluxe main table on its own. The supplemental files then arrived one at a time rather than all at once — multi-county from December 2007 (14 months after the Business edition itself), place FIPS not until April 2015, and the ACS file not until April 2022. All five are only in the bundle together from June 2023 onward. See what changes across the archive below, or the full data dictionary for every current column with its exact name and type.

What the archive covers — and what changes across it

A 21-year archive is not 21 years of the same file. The structure grew, and the Census underneath it turned over twice. None of that is a defect — it is the record of what was true at the time — but you should know it before you buy, not after.

Census data underneath

Field set

October 2004March 2026

Drawn to scale from the 257 publications currently in the archive. October 2004 is not just where our records start — it is the first ZIP Code Database we ever published. Nothing before it is missing, because there is nothing before it. The rightmost Census band grows every month; the two earlier Census boundaries are approximate, since that data was updated as soon after each release as we could process it rather than on a fixed date.

What varies from one end of the archive to the other

The field set

  • Business field set from October 2006 onward — that was our first Business publication.
  • The first 24 publications (October 2004 to September 2006) are the Deluxe field set — fewer columns, because the Business edition did not exist yet.
  • Planning a study that crosses 2006? Work from the Deluxe columns, which are present in every month of the archive.
  • Field names stay consistent throughout; it is the number of them that grows.

Supplemental files

  • They did not all arrive with the Business edition. Each was added when we built it, so an early Business month has the main table and little else.
  • Multi-county file: from December 2007 — 14 months after the Business edition itself.
  • 2010 Census by ZCTA: from October 2011.
  • Place FIPS file: from April 2015.
  • American Community Survey: from April 2022, starting with the 2016–2020 vintage. It is also the one file that never settles — it is re-cut each year as a new ACS is released, so every month carries whatever vintage was current then. Today's is 2020–2024.
  • 2020 Census by ZCTA: from June 2023, with the Census 2020 changeover.
  • All five together only exist from June 2023 — the last 34 of the 257 publications. Every earlier month has some subset. Tell us the month you need and we will confirm exactly which files it contains.

Census era

  • Each publication carries the best Census data available at the time.
  • Roughly 2004–2010: Census 2000 figures.
  • Roughly 2011–May 2023: Census 2010 figures.
  • June 2023 onward: Census 2020 figures.
  • A population change between two months can therefore be a Census turnover rather than a real move. Compare within an era.
  • Where the ACS file is present (April 2022 onward) those columns move faster still — a new vintage every year. Two such months a year apart are two different surveys, not a trend. Compare the same vintage, or expect the survey itself to be the change.

File formats

  • CSV, Excel and Access: available across the whole archive.
  • SQL creation scripts: roughly 2018 forward only — we did not publish them before that.
  • All formats within a month contain identical data.

What never changes

  • Same USPS-licensed sources as the current database, in every month.
  • Same monthly cadence, published on the 1st, with no gaps.
  • Frozen on release and never revised — which is the whole point for audit and compliance work.
  • One row per ZIP / city / alias combination throughout.

Found your era? Any month from October 2004 to March 2026 is available individually, or take all 257 at once.

Call for Details

What people actually use it for

Every one of these needs data as it existed on a date, which is the one thing a current database cannot give you.

Same standards, same sources. Historical publications are not a lesser product — they are this product, kept. Each one came from the same USPS-licensed sources and the same monthly process as the database we ship today. How we build it →

Recognize your problem above? Call 1-800-425-1169 and tell us the date or span you need.

Call for Details

Explore a live sample with our Data Explorer

These are live rows from the current database, shown in the Business structure that historical publications use. A month from the archive holds the same columns with the values that were true then. Filter to a state, then drill into a county.

Viewing: Business structure — the same shape as a historical month

Loading the live sample

Connecting to the current database to show the field structure a historical month uses.

What you'll see:

  • 94+ data fields per ZIP code
  • Geographic coordinates
  • Census demographics
  • Statistical classifications
  • Business and delivery detail

How it maps to a historical month:

  • Same columns, same names
  • Values are point-in-time, not current
  • Census figures follow the era
  • Early months carry fewer columns
  • Explore before you buy

Frequently asked questions

Ten questions we get about historical data, answered properly.

Time, and nothing else. A historical publication is the current Business database, frozen on the day it was published.

Current database:

  • Latest data, updated monthly on the 1st
  • Annual subscription with 12 monthly updates
  • Best for ongoing business operations
  • Three editions: Standard, Deluxe, Business

Historical database:

  • Point-in-time data from October 2004 to March 2026
  • One-time purchase, no updates — by design
  • Best for research, compliance and validation
  • Business structure from October 2006; Deluxe before that

Both come from the same USPS-licensed sources and the same monthly process.

October 2004 — which is also the first ZIP Code Database we ever published.

The archive does not begin where our record-keeping happened to start; it begins at the beginning. We kept a copy of the very first Deluxe publication and of every month since. That gives you:

  • 21 years of coverage
  • 257 monthly publications
  • Complete coverage through March 2026
  • Monthly granularity — no interpolation, no gaps

Historical availability runs a few months behind the current database, so that customers who need recent data buy the product built for it.

The archive stops a few months short of today so that the two products stay distinct.

The reasoning:

  • Historical exists for past research and compliance, not for recent data
  • Anyone who needs current data is better served by a subscription with monthly updates
  • It keeps customers from paying archive prices for something the current database already gives them

Which one do you need?

  • Data from the last few months? The current database, with monthly updates.
  • Data from a specific past date? A single historical month.
  • Change over years? The complete archive.

Still unsure? Contact us and we will tell you which one fits — including when the answer is the cheaper one.

A single month if:

  • You need one specific date for an audit or compliance response
  • You are validating addresses that existed on a particular date
  • You have a one-time project with a clear target date

The complete archive if:

  • You are looking at change across years rather than one date
  • You are not certain which month you will need
  • You expect to come back to historical data more than once

If there is any chance you will need a second date, the complete archive is usually the better answer.

Contact us if you would like help deciding.

Yes. It is not automated on the website, but we do it regularly — call 1-800-425-1169 or email [email protected] and we will work out the difference.

If you are uncertain up front, the complete archive removes the question entirely.

The same formats we published at the time:

  • CSV: universal compatibility, works with anything. Available across the whole archive.
  • Microsoft Excel (.XLS): available across the whole archive.
  • Microsoft Access (.MDB): available across the whole archive.
  • SQL scripts: roughly 2018 forward only — we did not produce them before that, so an older month will not have one.

Within any single month, every format contains identical data. Most customers take CSV.

Full format details — delimiters, encoding, line endings, keys — are on the technical specifications page.

Each publication carries the Census data that was current when it was published. The field names stay the same; the source underneath them turns over.

  • Census 2000 era — roughly 2004 to 2010.
  • Census 2010 era — roughly 2011 to May 2023. Updated as soon after release as we could process it.
  • Census 2020 era — June 2023 onward.

The practical warning: a population figure that jumps between two adjacent months may be recording a Census turnover, not a real change in the population. For trend work, compare within an era, or treat the boundary as a known discontinuity.

The coverage timeline above shows where those boundaries fall.

No, and that is the feature.

A historical publication is exactly what existed at a moment. If it changed, it would stop being useful for:

  • Compliance and audit responses
  • Validating addresses as they existed on a date
  • Research that has to be reproducible
  • Anything documented as "as of [date]"

Your purchase includes: a one-time payment, a year of download access, and no recurring fees.

Need ongoing updates? That is the current database with its monthly releases.

Yes — it is the single most common use.

Match your old records against the ZIP data from the month those records were created. That tells you whether the address was valid then, which is a different question from whether it is valid now.

A worked example: you hold customer records from 2010 whose addresses fail against the current file. Checked against the October 2010 publication, most of them pass — the records were fine, the ZIP boundaries moved. Now you know which ones are genuinely bad.

For ongoing, real-time validation, use the ZIP Code API instead.

Not of a specific past month, but you can check the structure three ways before spending anything:

  • The Data Explorer above — live rows in the same structure a historical month uses.
  • The Business sample file — download it here. Same shape, current values.
  • Ask us — tell us the month you are considering and we will confirm what that publication contains before you buy.

Every purchase is also covered by the 30-day money-back guarantee. If the data does not do what you needed, tell us within 30 days and we refund it.

Questions answered? Every purchase is covered for 30 days.

Call for Details

Not sure which month you need?

1-800-425-1169

Tell us the problem; we will tell you the month.

Email us

[email protected]

Answered within the hour during business hours.

Sources & references

Each publication contains the data as it existed on release. Sources include the United States Postal Service, the U.S. Census Bureau (Census 2000, 2010 and 2020 by era), the U.S. Geological Survey, the Centers for Medicare & Medicaid Services and other federal agencies. Every monthly file in the archive is the file we delivered to customers that month — which is what makes it usable as evidence.

  1. U.S. Postal Service [Historical publications: October 2004 – March 2026], https://www.usps.com
  2. ZIP-Codes.com proprietary data enhancements [Historical archive], https://www.zip-codes.com
  3. U.S. Census Bureau [Multiple eras], https://www.census.gov
    • Census 2000 [Decennial] — used in early publications (approx. 2004–2010)
    • Census 2010 [Decennial] — used in mid-period publications (approx. 2011–2023)
    • Census 2020 Demographic & Housing Characteristics (DHC) [Decennial] — used from June 2023
    • American Community Survey (ACS) [Annual] — various years as released
    • U.S. Census Business Patterns (CBP) [Annual] — various years as released
    • American National Standards Institute (ANSI) [As released]
    • Office of Management and Budget (OMB) [As released]
  4. U.S. Geological Survey [As released], https://www.usgs.gov
  5. Centers for Medicare & Medicaid Services [Annual], https://www.cms.gov
  6. National Center for Education Statistics [As released], https://nces.ed.gov
Portions of data are provided by the United States Postal Service and are © 2004–2026. Zip-Codes.com is a non-exclusive licensee of the United States Postal Service.