This study was to investigate the effects of balance exercises on among static, dynamic and reactive balance ability in community-dwelling older women. Participants comprised 26 community-dwelling older women in Japan, randomized for age and gait speed to the exercise group (n=13) or the control group (n=13). The exercise group performed structured balancing exercises on the sponge and the G ball, once weekly for 24 weeks. The control group performed stretching exercises once per month. Static balance ability was assessed by measuring postural sway and standing on one leg; dynamic balance ability was assessed by measuring functional reach, timed up and go, and gait; and reactive balance ability was assessed using the EquiTest. We also measured knee and ankle strength using an isokinetic machine.At baseline, the two groups were well matched in physical characteristics and in all balance ability and strength tests. After 24 weeks, step length of preferred speed walking (p=0.049) only improved, and peak torque of knee flexion (p=0.050) tended to improve in the exercise group. Any correlations were not found between step length of preferred speed walking and knee flexion strength.These results suggest that these structured balance exercises are effective in improving dynamic, but not static or reactive balance ability. Improvements in dynamic balance ability were not dependent on improved muscle strength.