In general, differentiating between various 8-bit character sets is a hairy problem. If you have nothing else to go on besides the text files, I suspect you're going to need clever heuristics. But try Encode::Guess first; it might be enough.