The present study examined whether the degree to which muscle strength is improved by a body mass-based home exercise program in middle-aged and older women depends on the force-generating capabilities of the muscles prior to the intervention. 75 women (53-76 years) voluntarily participated in a circuit training program consisting of 5 exercises (16 repetitions/exercise, 2 or 3 circuits/day) using only body mass as resistance for 3 months. The subjects performed the training program 6 days a week in their own home and once a week in a local gym. Before and after intervention, isometric torques during maximal voluntary knee extension (KET) and plantar flexion (PFT) were determined and expressed relative to body mass (KET/BM and PFT/BM, respectively). KET/BM and PFT/BM increased significantly after intervention, and their relative changes were negatively correlated to the absolute values before intervention. Most of the subjects whose KET/BM and PFT/BM values before intervention were greater than 2.8 Nm/kg and 1.7 Nm/kg, respectively, did not show increases in strength after intervention. Thus, although body mass-based exercise at home is effective in improving lower limb strength in middle-aged and older women, the magnitude of the improvement is influenced by the force-generating capability before intervention.