EBCDIC is an extension to 8 bits of BCDIC (Binary Coded Decimal Interchange Code), an earlier 6-bit character set used on IBM computers. EBCDIC was [first?] used on the successful System/360, anounced on 1964-04-07, and survived for many years despite the almost universal adoption of ASCII elsewhere. Was this concern for backward compatibility or, as many believe, a marketing strategy to lock in IBM customers?
IBM created 57 national EBCDIC character sets and an International Reference Version (IRV) based on ISO 646 (and hence ASCII compatible). Documentation on these was not easily accessible making international exchange of data even between IBM mainframes a tricky task.
US EBCDIC uses more or less the same characters as ASCII, but different code points. It has non-contiguous letter sequences, some ASCII characters do not exist in EBCDIC (e.g. square brackets), and EBCDIC has some (cent sign, not sign) not in ASCII. As a consequence, the translation between ASCII and EBCDIC was never officially completely defined. Users defined one translation which resulted in a so-called de-facto EBCDIC containing all the characters of ASCII, that all ASCII-related programs use.
Some printers, telex machines, and even electronic cash registers can speak EBCDIC, but only so they can converse with IBM mainframes.
For an in-depth discussion of character code sets, and full translation tables, see Guidelines on 8-bit character codes (ftp://ftp.ulg.ac.be/pub/docs/iso8859/iso8859.networking).