To evaluate the efficacy of therapeutic exercises in the prevention of bone loss, 146 untrained healthy postmenopausal women were prospectively controlled for (mean +/- SD) 3.0 +/- 1.3 yr. Eighty-two subjects aged (mean +/- SD) 61.5 +/- 6.1 yr participated in an exercise program (group 1) and sixty-four aged (mean +/- SD) 59.1 +/- 7.4 yr served as controls (group 2). Periodically during the study period, we measured women's bone density at two forearm sites and recorded their physical activities. Because bone loss differed insignificantly between the groups, group 1 was retrospectively subdivided into group 1a (regular exercise) and group 1b (nonregular exercise). The results showed that only 39 women (48 percent) of group 1 (group 1a) performed the exercise program regularly for the prescribed time. Regression slopes of forearm bone density (distal and proximal scans) v time were significantly less negative (P < 0.05) in group 1a (distal, -0.3 percent and proximal, -0.7 percent per year) than in group 1b (distal, -1.8 percent and proximal -1.6 percent per year) or group 2 (distal, -1.7 percent and proximal, -1.9 percent per year). We conclude that in untrained elderly women, poor compliance with regular physical activities is a main factor, explaining the lack of response to exercise treatment in prevention of osteoporosis.