Every row in every table is automatically assigned a row identifier. Row IDs are 64-bit unsigned integers. Each row has a row ID unique within the containing table. The ID of a row is constant and never changes. The row ID is not part of the table’s schema, and is not considered to be a column of the table. However, it can be queried and used within SQL commands as described below.

Row IDs can be returned from queries instead of table columns or expressions. For example: SELECT ROWID FROM my_table. See the SELECT command for more details.

Row IDs can also be used in the WHERE clause of commands.

Row IDs of the affected rows are returned by the INSERT, UPDATE and DELETE commands if the RETURNING ROWID clause is specified.