utf8_decode() and mixed character sets

Hey everyone. I'd been troubled for a while by the fact that inserting cut-pasted special characters such as ä caused truncation when passed to MySQL, then discovered that it was because I was cutting and pasting unicode values into non-unicode Latin-1 strings. Since Latin-1 also has equivalent values, I was hoping that filtering my mixed unicode/non-unicode string through utf8_decode() would solve the problem, but instead, where the unicode character used to be, I now get a '?', followed by a few characters being taken out of the middle. I'm guessing that this is because utf8_decode() assumes the whole string is unicode and therefore removes a bunch of extra bytes from the string and corrupts it. At least, that's my guess. I could be very wrong (I have pretty much no experience with different character sets...) My question is, what's a good way to translate unicode characters in a non-unicode string to their Latin-1 equivalents? I need to be able to do this in order to sanitize a fairly common form of input. Thanks! James