The more you know about the data, the better. For instance, if most strings will not match, you could do some preprocessing: for each string, split it on whitespace. Have a hash of all the first words of the keys of %hash; if none of the words in the string are in the hash, there cannot be a match. Obviously, this will not do you any good if all strings will have a match. Also, is it important to check the keys in a specific order? Or are you satisfied if any key matches? And why do you have %hash if you aren't using its values?