Normally this is a misleading error - one of the *first* things we do when runnign import is try derive metadata (like the characterset) from the header information in the file.

If the file is corrupted, or from the wrong version of exp, or was done with expdp etc...ie, anything that makes the file no longer suitable, then we can't find the characterset info and report this error.

So double check your file, and where it was sourced from.

Hope this helps.