Skip to content
PostgreSQL Time Zone Abbreviations Explained

Click to use (opens in a new tab)

PostgreSQL Time Zone Abbreviations Explained

November 16, 2026 by Chat2DBChat2DB Team

A timestamp string such as '2027-01-15 09:00 CST' looks unambiguous until you remember that CST is Central Standard Time in North America, China Standard Time, and Cuba Standard Time, three different offsets. IST is Israel, India or Ireland. PostgreSQL has to pick one, and how it picks depends on a configuration parameter, on the session's time zone, and since PostgreSQL 18 on a new lookup order.

Abbreviations are in the news again because of the November 2026 tzdata update: Manitoba's new year-round time is labelled EST, and Inuvik's is labelled CST. For full zone names that PostgreSQL rejects, see time zone not recognized. This guide explains how PostgreSQL resolves abbreviations in input, how that differs from output, how to customize the table, and the errors you can hit. All examples were run on PostgreSQL 18.7 and 17.12.

Abbreviations Are Not Time Zones

PostgreSQL treats three kinds of time zone specifications differently:

  • Full zone names such as America/Chicago carry a complete history of offsets and daylight saving rules. They are what you should use for SET timezone and for converting with AT TIME ZONE.
  • Abbreviations such as CST stand for one fixed offset. CST always means UTC-6 in the default table, even in July when Chicago is on CDT.
  • Numeric offsets such as -06 or +05:30 are fixed offsets written explicitly.

Abbreviations are recognized in input strings for timestamptz and timetz, and in AT TIME ZONE (where AT TIME ZONE 'CST' applies a fixed UTC-6). They are not accepted as a TimeZone setting: SET timezone = 'CST' fails with invalid value for parameter "TimeZone": "CST". SET timezone = 'EST' works only because tzdata also contains a zone that happens to be named EST. PostgreSQL looks abbreviations up in a table chosen by the timezone_abbreviations parameter, whose default value is Default.

What the Default Table Says

The pg_timezone_abbrevs view lists the abbreviations PostgreSQL currently accepts:

SET timezone = 'UTC';
SELECT abbrev, utc_offset, is_dst
FROM pg_timezone_abbrevs
WHERE abbrev IN ('EST', 'CST', 'IST', 'BST', 'AEST', 'SGT', 'KST')
ORDER BY abbrev;
 abbrev | utc_offset | is_dst
--------+------------+--------
 AEST   | 10:00:00   | f
 BST    | 01:00:00   | t
 CST    | -06:00:00  | f
 EST    | -05:00:00  | f
 IST    | 02:00:00   | f
 KST    | 09:00:00   | f
 SGT    | 08:00:00   | f

With the session in UTC, the Default table resolves the ambiguous ones like this:

SELECT '2027-01-15 09:00 EST'::timestamptz AS est,
       '2027-01-15 09:00 CST'::timestamptz AS cst,
       '2027-01-15 09:00 IST'::timestamptz AS ist,
       '2027-01-15 09:00 BST'::timestamptz AS bst;
          est           |          cst           |          ist           |          bst
------------------------+------------------------+------------------------+------------------------
 2027-01-15 14:00:00+00 | 2027-01-15 15:00:00+00 | 2027-01-15 07:00:00+00 | 2027-01-15 08:00:00+00

IST became UTC+2 (Israel Standard Time) and CST became UTC-6 (North American Central). The Default file itself documents the conflicts with comments such as # CONFLICT! IST is not unique, listing Irish and Indian Standard Time as the losers. The full view in our build listed 195 abbreviations.

The PostgreSQL 18 Change: Session Zone First

PostgreSQL 18 changed the lookup order. In the release notes' words, the system "will now favor the current session's time zone abbreviations before checking the server variable timezone_abbreviations". If the session's TimeZone uses an abbreviation (in tzdata, at any point in its history), that meaning wins.

The same queries on 17.12 and 18.7:

SET timezone = 'Asia/Kolkata';
SELECT '2027-01-15 09:00 IST'::timestamptz;
 
SET timezone = 'Asia/Shanghai';
SELECT '2027-01-15 09:00 CST'::timestamptz;
 
SET timezone = 'America/Havana';
SELECT '2027-01-15 09:00 CST'::timestamptz;
 
SET timezone = 'Europe/Dublin';
SELECT '2027-07-15 09:00 IST'::timestamptz;
Session zoneInputPostgreSQL 17.12PostgreSQL 18.7
Asia/Kolkata09:00 IST12:30:00+05:30 (read as UTC+2)09:00:00+05:30
Asia/Shanghai09:00 CST23:00:00+08 (read as UTC-6)09:00:00+08
America/Havana09:00 CST11:00:00-04 (read as UTC-6)10:00:00-04 (read as UTC-5)
Europe/Dublin09:00 IST08:00:00+01 (read as UTC+2)09:00:00+01

On 17, an Indian user typing 09:00 IST got a time three and a half hours off. On 18, the session zone resolves it the way the user meant. The rule only helps when the session zone actually uses that abbreviation: with Australia/Adelaide, CST still resolves through the default table to UTC-6 on 18.7, because tzdata labels Adelaide time ACST, so '2027-07-15 09:00 CST' lands at 2027-07-16 00:30:00+09:30.

This change can alter results after a major upgrade. If an application in a session set to Asia/Kolkata relied on IST meaning UTC+2 (unlikely, but possible in data feeds from Israel processed on Indian servers), the same input now yields a different instant. Treat abbreviation-bearing inputs as something to retest when you upgrade to 18.

Abbreviations in AT TIME ZONE

Because an abbreviation stands for one fixed offset, using it in AT TIME ZONE gives a different answer from the zone it is named after whenever daylight time is in effect:

SET timezone = 'UTC';
SELECT ts AT TIME ZONE 'CST'             AS fixed_cst,
       ts AT TIME ZONE 'America/Chicago' AS chicago,
       ts AT TIME ZONE 'CDT'             AS fixed_cdt
FROM (VALUES (timestamptz '2027-07-15 15:00+00'), (timestamptz '2027-01-15 15:00+00')) v(ts);
      fixed_cst      |       chicago       |      fixed_cdt
---------------------+---------------------+---------------------
 2027-07-15 09:00:00 | 2027-07-15 10:00:00 | 2027-07-15 10:00:00
 2027-01-15 09:00:00 | 2027-01-15 09:00:00 | 2027-01-15 10:00:00

In July, CST is an hour behind what clocks in Chicago show; in January, CDT is an hour ahead. Reports that group by local day or hour with an abbreviation are therefore wrong for half the year. Use the full zone name whenever the intent is "local time in that place".

Errors You Will See

An abbreviation that is not in the table and not used by the session zone makes the whole literal invalid. The error does not mention time zones:

SET timezone = 'UTC';
SELECT '2027-01-15 09:00 WIB'::timestamptz;
ERROR:  invalid input syntax for type timestamp with time zone: "2027-01-15 09:00 WIB"
LINE 1: SELECT '2027-01-15 09:00 WIB'::timestamptz;
               ^

WIB (Western Indonesia Time) is not in Default. A misspelled abbreviation gives exactly the same error, so if a CSV import fails on timestamps, check the abbreviations before the date format.

Setting the parameter to a file that does not exist fails immediately:

SET timezone_abbreviations = 'Nope';
ERROR:  invalid value for parameter "timezone_abbreviations": "Nope"

(Note the different message from a bad zone name in AT TIME ZONE, which is time zone "..." not recognized.)

Choosing a Different Table

PostgreSQL ships several abbreviation files in share/timezonesets (find the directory with pg_config --sharedir). Besides Default there are Australia and India, which resolve the conflicting abbreviations the regional way, plus .txt files per region that the main files include. Switching to India:

SET timezone = 'UTC';
SET timezone_abbreviations = 'India';
SELECT '2027-01-15 09:00 IST'::timestamptz AS ist_india;
       ist_india
------------------------
 2027-01-15 03:30:00+00

You can set the parameter per session, per role, per database or in postgresql.conf:

ALTER DATABASE app_in SET timezone_abbreviations = 'India';

Writing a Custom File

When none of the shipped files fits, create your own in share/timezonesets. The format is one abbreviation per line with an offset in seconds east of UTC, and an optional D for daylight time. @INCLUDE pulls in another file, and @OVERRIDE allows later lines to replace earlier ones:

@INCLUDE Default
@OVERRIDE
# Indonesia (not in Default)
WIB 25200    # Western Indonesia Time, UTC+7
WITA 28800   # Central Indonesia Time, UTC+8
# make IST mean India instead of Israel
IST 19800    # India Standard Time, UTC+5:30

Saved as MyApp:

SET timezone = 'UTC';
SET timezone_abbreviations = 'MyApp';
SELECT '2027-01-15 09:00 WIB'::timestamptz AS wib,
       '2027-01-15 09:00 IST'::timestamptz AS ist;
          wib           |          ist
------------------------+------------------------
 2027-01-15 02:00:00+00 | 2027-01-15 03:30:00+00

Without the @OVERRIDE line, redefining IST is an error, and the message says exactly where the conflict is:

ERROR:  time zone abbreviation "ist" is multiply defined
DETAIL:  Entry in time zone file "Default", line 244, conflicts with entry in file "Broken", line 3.

The file lives inside the PostgreSQL installation, so package upgrades and new servers will not have it unless you deploy it with your configuration management. Put it in the same place as postgresql.conf changes in your provisioning code.

Abbreviations in Output

Output works the other way around. When PostgreSQL prints a timestamptz with to_char(..., 'TZ') or in the default DateStyle, it uses the abbreviation that tzdata defines for the session zone at that instant, not the timezone_abbreviations table:

SET timezone = 'America/Winnipeg';
SELECT to_char(timestamptz '2027-01-15 15:00+00', 'HH24:MI TZ') AS winnipeg,
       '2027-01-15 09:00 EST'::timestamptz AS est_in_winnipeg;
 winnipeg  |    est_in_winnipeg
-----------+------------------------
 10:00 EST | 2027-01-15 09:00:00-05

Since tzdata 2026f, Winnipeg prints EST in winter, and EST input means UTC-5 both in the default table and in the Winnipeg zone, so the round trip is consistent. Many zones no longer have letter abbreviations at all, because tzdata removed invented ones. São Paulo prints a numeric abbreviation:

SET timezone = 'America/Sao_Paulo';
SELECT to_char(timestamptz '2027-01-15 15:00+00', 'HH24:MI TZ OF') AS sp;
      sp
---------------
 12:00 -03 -03

Numeric abbreviations like -03 are valid input as well, and they are the safest form to exchange.

Recommendations

  1. Exchange offsets, not abbreviations. ISO 8601 with a numeric offset (2027-01-15T09:00:00-05:00) or UTC (Z) has exactly one meaning everywhere.
  2. Store zone names, not abbreviations, when you need to know where a user is. America/Winnipeg survives rule changes; CST does not even say whether daylight time applies.
  3. Retest abbreviation inputs when upgrading to PostgreSQL 18, especially in sessions whose zone uses an abbreviation that conflicts with the default table (India, China, Cuba, Ireland).
  4. If you must accept abbreviations from a fixed region, set timezone_abbreviations per database or role, or ship a custom file with @OVERRIDE, and deploy it with your server configuration.
  5. Use pg_timezone_abbrevs to answer "what does PostgreSQL think XYZ means right now" instead of guessing.

When you are cleaning up a table full of mixed timestamp strings, it helps to test candidate parsing rules interactively against real rows. Chat2DB (opens in a new tab) can run the same conversion query with different timezone_abbreviations settings in separate tabs, so you can compare which rows change before committing to one.

Summary

PostgreSQL resolves abbreviations in timestamp input through the timezone_abbreviations table (Default unless changed), and since PostgreSQL 18 it first checks the abbreviations used by the session's own time zone, which fixes IST in India, CST in China and Cuba, and IST in Ireland. Unknown abbreviations produce a generic invalid input syntax error. You can switch to the shipped India or Australia tables or write a custom file with @INCLUDE and @OVERRIDE. Output abbreviations come from tzdata, which is why Winnipeg now prints EST. For anything that crosses system boundaries, numeric offsets and full zone names are the reliable choice.