PostgreSQL Data Types Reference

How to use

1. 1Type a PostgreSQL type name, family, or alias into the search box (try text, varchar, numeric, timestamptz, jsonb, uuid, int4, money).
2. 2Click a family chip (Numeric, Character, Date and time, JSON, UUID, Network, Container) to filter the catalog.
3. 3Click any entry card to pin its detail view, which shows storage, value range, a DDL column example, an INSERT example, and the readback shape.
4. 4Read the notes section for PostgreSQL-specific gotchas (timestamp vs timestamptz, varchar vs text, json vs jsonb, serial vs identity).
5. 5Use Copy entry to grab a markdown-formatted block for a runbook or pull request comment, or Copy full reference to export the whole cheat sheet at once.

About this tool

PostgreSQL Data Types Reference is a searchable, copy-ready cheat sheet for every PostgreSQL data type a developer reaches for when writing DDL, designing a schema, or tuning column widths in EXPLAIN. Entries cover the eleven families from the official docs: numeric (smallint, integer, bigint, numeric, real, double precision, the serial helpers, and money), character (text, varchar, char, citext from contrib), date and time (date, time, timestamp, timestamptz, interval), boolean, uuid, binary (bytea), network address (inet, cidr, macaddr), JSON (jsonb and json), geometric (point and the planar family), container (arrays, numeric and date and time range types), and the special system types (oid, pg_lsn). Each entry carries the exact storage footprint (smallint = 2 bytes fixed, numeric = around 5 bytes plus 2 bytes per group of four digits, bytea = variable up to 1 GB), the value range, a one-paragraph description that captures the PostgreSQL-specific behavior a developer actually cares about (TOAST limits, alignment padding, timezone semantics, JSONB key normalization, default 1-based array indexing), a worked DDL column declaration ready to drop into a CREATE TABLE statement, an INSERT example that exercises the type, and the way the value reads back from the server. A search box scores matches across the type name, family, aliases (int4, int8, character varying, double precision), and use cases, so a query like 'currency' surfaces numeric and money side by side. Filters group results into the eleven canonical families. Each detail card includes a copy button for the entry, the DDL, the insert, the readback, plus a 'related types' row that links to the obvious siblings (smallint and bigint next to integer, jsonb next to json, timestamptz next to timestamp). Notes call out the gotchas seasoned PostgreSQL engineers wish they had read sooner: prefer GENERATED ALWAYS AS IDENTITY over serial, varchar(n) and text are storage-identical so reach for text by default, money formatting depends on lc_monetary and is rarely what an application wants, json preserves bytes while jsonb normalizes them, timestamptz stores UTC and re-renders in the session zone, and arrays make a poor stand-in for a child table when you need referential integrity. Useful for backend engineers writing migrations, data platform teams designing a fact table, database administrators sizing storage, students working through the PostgreSQL tutorial, and code reviewers double-checking a column type before approving a pull request. Everything ships pre-baked and runs locally in your browser. No upload, no remote dictionary, no signup.

Free to use. Works in your browser. No signup, no login.