Self-efficacy is increasingly used as a predictor of health behavior. The purpose of this study was to examine the impact of exercise self-efficacy on exercise behaviors and outcomes. A one-group pre-test/post-test design was used. The treatment, a 12-week exercise training program, was executed between the pre- and post-tests. Exercise self-efficacy was measured prior to training and at the 4th, 8th and 12th weeks of training. Estimated VO2max, fatigue, anxiety, depression, and quality of life (QOL) were assessed prior to exercise training and after 12 weeks of training. Compliance rate and exercise intensity were computed at the 4th, 8th and 12th weeks of training. Results of this study revealed no relationship between exercise self-efficacy and compliance rate, nor between exercise self-efficacy and exercise intensity. The change in exercise self-efficacy after exercise training, rather than the initial self-efficacy level, was significantly related to exercise outcomes. Exercise intensity was more important in predicting the improvement of VO2max than was compliance rate. In contrast, compliance rate was more important in predicting the improvement of QOL than was exercise intensity.