MySQL get missing IDs from table

To add a little to Ivan's answer, this version shows numbers missing at the beginning if 1 doesn't exist:

SELECT 1 as gap_starts_at,
       (SELECT MIN(t4.id) -1 FROM testtable t4 WHERE t4.id > 1) as gap_ends_at
FROM testtable t5
WHERE NOT EXISTS (SELECT t6.id FROM testtable t6 WHERE t6.id = 1)
HAVING gap_ends_at IS NOT NULL limit 1
UNION
SELECT (t1.id + 1) as gap_starts_at, 
       (SELECT MIN(t3.id) -1 FROM testtable t3 WHERE t3.id > t1.id) as gap_ends_at
FROM testtable t1
WHERE NOT EXISTS (SELECT t2.id FROM testtable t2 WHERE t2.id = t1.id + 1)
HAVING gap_ends_at IS NOT NULL;

A more efficent query:

SELECT (t1.id + 1) as gap_starts_at, 
       (SELECT MIN(t3.id) -1 FROM my_table t3 WHERE t3.id > t1.id) as gap_ends_at
FROM my_table t1
WHERE NOT EXISTS (SELECT t2.id FROM my_table t2 WHERE t2.id = t1.id + 1)
HAVING gap_ends_at IS NOT NULL

SELECT a.id+1 AS start, MIN(b.id) - 1 AS end
    FROM testtable AS a, testtable AS b
    WHERE a.id < b.id
    GROUP BY a.id
    HAVING start < MIN(b.id)

Hope this link also helps http://www.codediesel.com/mysql/sequence-gaps-in-mysql/

Rather than returning multiple ranges of IDs, if you instead want to retrieve every single missing ID itself, each one on its own row, you could do the following:

SELECT id+1 FROM table WHERE id NOT IN (SELECT id-1 FROM table) ORDER BY 1

The query is very efficient. However, it also includes one extra row on the end, which is equal to the highest ID number, plus 1. This last row can be ignored in your server script, by checking for the number of rows returned (mysqli_num_rows), and then using a for loop if the number of rows is greater than 1 (the query will always return at least one row).

Edit: I recently discovered that my original solution did not return all ID numbers that are missing, in cases where missing numbers are contiguous (i.e. right next to each other). However, the query is still useful in working out whether or not there are numbers missing at all, very quickly, and would be a time saver when used in conjunction with hagensoft's query (top answer). In other words, this query could be run first to test for missing IDs. If anything is found, then hagensoft's query could be run immediately afterwards to help identify the exact IDs that are missing (no time saved, but not much slower at all). If nothing is found, then a considerable amount of time is potentially saved, as hagensoft's query would not need to be run.