---
title: "citext"
slug: "citext"
category: "glossary"
tags: ["glossary", "database", "postgres", "citext", "text", "case-insensitive", "sql"]
status: "stable"
last_updated: 2026-08-14
summary: "citext is a PostgreSQL extension type for case-insensitive text storage and comparison, removing the need for lower() wrappers in queries."
related:
  [
    "[[glossary/btree-index]]",
    "[[glossary/query-plan]]",
    "[[glossary/gin-index]]",
    "[[backend/postgres]]",
    "[[glossary/sequential-scan]]",
  ]
---

## Overview

This page is the atomic definition. Postgres extension and schema guidance lives at [[backend/postgres]]. `citext` is a PostgreSQL extension type that stores text with its original case but compares and sorts it case-insensitively, replacing manual `lower()` wrapping.

## Definition

`citext` (case-insensitive text) is a PostgreSQL extension data type in `contrib/citext`. It stores the original case but compares and sorts case-insensitively using `lower()` semantics. Equality, `LIKE`, `ILIKE`, `ORDER BY`, and unique constraints all operate case-insensitively.

Without `citext`, case-insensitive lookups require a function-based index:

```sql
SELECT * FROM users WHERE lower(email) = lower($1);
CREATE INDEX idx_users_email_lower ON users (lower(email));
```

With `citext`, a plain B-tree index works and the query needs no `lower()`:

```sql
SELECT * FROM users WHERE email = $1; -- 'user@EXAMPLE.com' = 'user@example.com'
CREATE INDEX idx_users_email ON users (email);
```

`citext` comparisons call `lower()` internally on both sides. This is correct for ASCII and most Latin scripts but may not behave as expected for Turkish or Greek locale-sensitive case folding. For locale-aware case folding, use ICU collations instead.

Unique constraints on `citext` columns enforce uniqueness case-insensitively.

## When it applies

Use `citext` for email addresses, usernames, slugs, and any column where case-insensitive equality or uniqueness is required. Install the extension once per database with `CREATE EXTENSION citext`. Avoid `citext` when you need to preserve and compare the original case (e.g., passwords, tokens, code identifiers).

## Example

```sql
CREATE EXTENSION IF NOT EXISTS citext;

CREATE TABLE users (
  id       bigint PRIMARY KEY GENERATED ALWAYS AS IDENTITY,
  email    citext NOT NULL,
  username citext NOT NULL,
  UNIQUE (email),
  UNIQUE (username)
);

INSERT INTO users (email, username) VALUES ('Alice@Example.com', 'Alice');
INSERT INTO users (email, username) VALUES ('alice@example.com', 'alice'); -- ERROR: unique violation

SELECT * FROM users WHERE email = 'ALICE@EXAMPLE.COM'; -- returns row
```

## Related concepts

- [[glossary/btree-index]] - standard B-tree indexes work on `citext` without function wrapping.
- [[glossary/query-plan]] - `citext` equality uses a standard index scan; no special planning needed.
- [[glossary/gin-index]] - use GIN on `citext` for full-text or ILIKE pattern queries.
- [[backend/postgres]] - the Postgres deep-dive.

## Citing this term

> See [[glossary/citext|citext]] (llmbestpractices.com/glossary/citext).

## Related

- [[glossary/btree-index]]
- [[glossary/query-plan]]
- [[glossary/gin-index]]
- [[backend/postgres]]
- [[glossary/sequential-scan]]