This randomized controlled trial with blinded assessment aimed to determine the effect of a 6-month minimally supervised exercise program on fall risk factors in people with Parkinson's disease (PD). Forty-eight participants with PD who had fallen or were at risk of falling were randomized into exercise or control groups. The exercise group attended a monthly exercise class and exercised at home three times weekly. The intervention targeted leg muscle strength, balance, and freezing. The primary outcome measure was a PD falls risk score. The exercise group had no major adverse events and showed a greater improvement than the control group in the falls risk score, which was not statistically significant (between group mean difference = -7%, 95% CI -20 to 5, P = 0.26). There were statistically significant improvements in the exercise group compared with the control group for two secondary outcomes: Freezing of Gait Questionnaire (P = 0.03) and timed sit-to-stand (P = 0.03). There were statistically nonsignificant trends toward greater improvements in the exercise group for measures of muscle strength, walking, and fear of falling, but not for the measures of standing balance. Further investigation of the impact of exercise on falls in people with PD is warranted.