Cursors in SQL procedures allow defining a result set that can be iterated through row by row. A cursor acts as a pointer to each row in turn. To use a cursor, it must be declared to define the result set, opened to establish the set, individual rows can then be fetched and processed one at a time using variables, and the cursor is closed once complete. Basic cursor usage involves the DECLARE, OPEN, FETCH and CLOSE statements. An example demonstrates summing the salaries from an employee table by declaring a cursor over it, fetching rows into a variable and accumulating the sum in a loop.