Ah, didn't think of replace. Certainly that can work if the domain of special characters is just the one character, and it never appears within the strings. If it could appear as part of one of the strings in the middle, like this: 12§345§ , then that won't work.

The domain is the possible set of values. Is this special character always the only one? Could there be some other character, or no? It almost seems like this might be a character not in your code page, which would make me worry that some other value might appear, or the underlying data contains different values.

Your LIKE doesn't work because it's matching essentially everything that has alphanumerics in the middle of the string.

I would still lean towards trying to strip off the last character if it matches this character, but if this is only at the end, Sean's solution will work.

The important thing here is to really think about the possible values of what will appear and not get too caught up in a small subset. Don't spin on this for days, but it's worth 20 minutes to really look at the data and possible values.