A term which describes a system that deals correctly with extended character sets which (unlike ASCII) use all eight bits of a byte. Many programs and communications systems assume that all characters have codes in the range to 127. This leaves the top bit of each byte free for use as a parity bit or some kind of flag bit. These assumptions break down when the program is used in some non-english-speaking countries with larger alphabets.
If a binary file is transmitted via a communications link which is not eight-bit clean, it will be corrupted. To combat this you can encode it with uuencode which uses only ASCII characters. There are some links however which are not even "seven-bit clean" and cause problems even for uuencoded data.