Twenty three geriatric patients from a mental institution were randomly assigned to three groups; a social group, an exercise group, and a control group. Subjects participated in the study for 12 weeks. Heart rate at rest, exercise and recovery, balancing ability, total daily activity level (diary method), and a self care inventory were recorded for analysis. Data were collected at the start, the 8th and 12th weeks of the study, and 4 weeks after the termination of treatment. A Groups X Sex X Measurement Period factorial design, using multivariate analysis of variance tested the effect of the three factors. The trials effect was significant, as were several univariate tests for the sex effect and group by sex interaction. Data were discussed in terms of programmed physical activity for geriatrics, its value and the limitations of evaluating such behavior in geriatric populations.