Implementation of a computer-based model for morphological analysis and synthesis of language, entitled P-KIMMO, is discussed. The model was implemented in Quintus Prolog on a Sun Workstation and exported to a Macintosh computer. This model has two levels of morphophonological representation, lexical and surface levels, associated by morphophonological rules that specify legitimate pairs of characters. The description offered here focuses on aspects of implementation only and not underlying theory. Components of the program are described, including structure of the lexicon, use of finite state automata to encode two-level rules, and the recognizer/generator algorithm. This version of the program is then compared and contrasted with a previously implemented version. Finally, procedures for use of the program on the UNIX and Mackintosh computers are outlined, with some screen illustrations. A brief bibliography is included, and a source listing of the P-KIMMO system is appended. (MSE)