Understanding the factors that influence adherence to exercise programs is necessary to develop effective interventions for people with cancer. We examined the predictors of adherence to a supervised exercise program for participants in the ENGAGE study - a cluster randomized controlled trial that assessed the efficacy of a clinician-referred 12-week exercise program among men treated for prostate cancer. Demographic, clinical, behavioral, and psychosocial data from 52 participants in the intervention group were collected at baseline through self-report and medical records. Adherence to the supervised exercise program was assessed through objective attendance records. Adherence to the supervised exercise program was 80.3%. In the univariate analyses, cancer-specific quality of life subscales (role functioning r = 0.37, P = 0.01; sexual activity r = 0.26, P = 0.06; fatigue r = -0.26, P = 0.06, and hormonal symptoms r = -0.31, P = 0.03) and education (d = -0.60, P = 0.011) were associated with adherence. In the subsequent multivariate analysis, role functioning (B = 0.309, P = 0.019) and hormonal symptoms (B = -0.483, P = 0.054) independently predicted adherence. Men who experienced more severe hormonal symptoms had lower levels of adherence to the exercise program. Those who experienced more positive perceptions of their ability to perform daily tasks and leisure activities had higher levels of adherence to the exercise program. Hormonal symptoms and role functioning need to be considered when conducting exercise programs for men who have been treated for prostate cancer.