A prefix code, also known as a prefix-free code or comma-free code, is a code constructed so that any partial code word, beginning at the start of a full code word but terminating prior to the end of that code word, is not itself a valid code word. In other words, for any given valid string in the code, there is no shorter string in the code that is an initial substring of that string.
This property permits the proper framing of transmitted code words when (a) external synchronization is provided to identify the start of the first code word in a sequence of code words and (b) no uncorrected errors occur in the symbol stream.
Examples of prefix codes are the variable-length Huffman codes, country calling codes, and ISBNs.
This article is partly derived from Federal Standard 1037C, which uses the term comma-free code.