BACKGROUND: Falls in the elderly are a major health problem. Although exercise programs have been shown to reduce the risk of falls, the optimal exercise components, as well as the working mechanisms that underlie the effectiveness of these programs, have not yet been established. OBJECTIVE: To test whether the Nijmegen Falls Prevention Program was effective in reducing falls and improving standing balance, balance confidence, and obstacle avoidance performance in community-dwelling elderly people. METHODS: A total of 113 elderly with a history of falls participated in this study (exercise group, n = 79; control group, n = 28; dropouts before randomization, n = 6). Exercise sessions were held twice weekly for 5 weeks. Pre- and post-intervention fall monitoring and quantitative motor control assessments were performed. The outcome measures were the number of falls, standing balance and obstacle avoidance performance, and balance confidence scores. RESULTS: The number of falls in the exercise group decreased by 46% (incidence rate ratio (IRR) 0.54, 95% confidence interval (CI) 0.36-0.79) compared to the number of falls during the baseline period and by 46% (IRR 0.54, 95% CI 0.34-0.86) compared to the control group. Obstacle avoidance success rates improved significantly more in the exercise group (on average 12%) compared to the control group (on average 6%). Quiet stance and weight-shifting measures did not show significant effects of exercise. The exercise group also had a 6% increase of balance confidence scores. CONCLUSION: The Nijmegen Falls Prevention Program was effective in reducing the incidence of falls in otherwise healthy elderly. There was no evidence of improved control of posture as a mechanism underlying this result. In contrast, an obstacle avoidance task indicated that subjects improved their performance. Laboratory obstacle avoidance tests may therefore be better instruments to evaluate future fall prevention studies than posturographic balance assessments.