WHERE keeps source rows whose condition is true. It applies before grouping in the logical reading of a SELECT query. Use it to narrow a query to the rows that answer your question, not as a way to repair a join that produced unexpected rows.

A basic filter

SELECT order_id, amount
FROM orders
WHERE status = 'paid';

Only rows with status = 'paid' are returned. The string uses single quotes because it is a value. A column name does not use single quotes. If status is NULL, the comparison is unknown, so that row is not returned.

Combine conditions carefully

SELECT order_id, amount
FROM orders
WHERE status = 'paid'
  AND amount >= 50;

Both conditions must be true. With OR, either condition can be true. Parentheses make mixed conditions clear:

SELECT order_id, amount
FROM orders
WHERE status = 'paid'
  AND (amount >= 50 OR customer_id = 10);

Without parentheses, SQL operator precedence might produce a result you did not intend. Write the grouping explicitly when it matters.

Ranges and missing values

For dates and timestamps, a half-open range is often safer than trying to include the last possible instant of a day:

SELECT order_id
FROM orders
WHERE placed_at >= '2026-01-01'
  AND placed_at < '2026-02-01';

This includes all of January for timestamps interpreted in the same chosen time zone. Decide which time zone the date boundaries mean before using this query in an application.

Use IS NULL and IS NOT NULL for missing values:

SELECT order_id
FROM orders
WHERE shipped_at IS NULL;

shipped_at = NULL does not test for missing values. A comparison with NULL produces unknown.

Values from Python or Java

When a value comes from a user or another system, pass it as a query parameter. Do not build WHERE customer_id = ... by joining raw input into a SQL string. Python database drivers and Java prepared statements use different placeholder syntax, but both keep the value separate from SQL code.

Check your understanding

  1. What rows does WHERE status = 'paid' AND amount >= 50 keep?
  2. Why does WHERE shipped_at = NULL fail to find unshipped orders?
  3. What time zone does a date boundary use when placed_at is timestamptz?

Further reading