This study replicated a social skills training program previously designed to eliminate verbally aggressive behavior in six institutionalized elderly, by substituting physical acts of aggression for verbal ones. The treatment package consisted of instructions, modeling, role playing, and feedback. Dependent measures included confirmed incidents of physically aggressive behavior monitored across an ABAB design with a 5-month follow-up period. Results indicated that physically aggressive behavior could be significantly decreased in a group training setting and subsequently generalized to ward and other socialized behavior. Ancillary aspects of the study include the dichotomy between motor and language aggression, individual preferences as they relate to intentionally aggressive acts and the profound need for replication studies to ensure external validity by minimizing error variation. Implications for future research are discussed, and references are included. (Author/TE)