Ok, let's dissect this:

NOT LIKE '%[^a-zA-Z]%'

The regexp here is "does not contain any letters". If we negate using NOT LIKE , it'll mean "contains only letters".

Now:

name LIKE '%[A-Za-z]%'

Regexp here means "contains at least one letter". So LIKE on that will mean the same thing.

So:

SELECT * FROM (VALUES ('A123'), ('123'), ('AAA'), (',')) x(name)
WHERE name LIKE'%[A-Za-z]%'

This will return both A123 (there's an A) and AAA , but neither 123 (only digits) nor , (still not a letter) will get returned.

SELECT * FROM (VALUES ('A123'), ('123'), ('AAA'), (',')) x(name)
WHERE name NOT LIKE '%[^A-Za-z]%'

This one will only return AAA because it's the only one containing letters only. AAA is the only record that regexp will not match, and since we're negating with NOT LIKE , it means only this one meets the WHERE condition.

This expression:

WHERE name NOT LIKE '%[^a-zA-Z]%'

matches all names that have a character that is not a letter, such as . or ; or whatever.

This expression:

name LIKE '%[a-zA-Z]'

matches all names that contain at least one letter.

So, 'ABC.DEF' will match both patterns -- it matches the first because it has . and the second because it has a letter.

However, 'ABC' does not match the first pattern because it only consists of letters. It does match the second one.

Following condition will check whether your column value has any string character from a-z or A-Z in it.

where name NOT REGEXP '[a-zA-Z]'