It true that when not in string we ca look for the next interesting char with a regexp like '\'|"|`|#|; and then decide what to do with a select/case. Well... I'm not sure that this will bring better performances... usually, there is not much characters between an end of string and the next interesting character. Will it really be faster ? If you have big text strings with few quotes in it, yes. But before using ereg(), you have to do some substr() (which can cost much time), don't you ? Well. If you want to test such an idea, you can. In my opinion substr() has quite a heavy cost... Benjamin