The current study examined exercise self-efficacy and exercise behavior during and after a strength-training intervention program with older adults. A model with cross-lagged and contemporaneous paths was tested with structural equations. Within testing occasions, higher physical resistance was related to greater beliefs in efficacy and control over exercise. At 3 months into the intervention, those who had higher physical resistance were less likely to show subsequent changes in beliefs. Those who had higher self-efficacy and control beliefs at 6 months were more likely to report that they were still exercising at 9 and 12 months after the intervention. Findings indicate that exercise self-efficacy and exercise behavior are associated with one another and that beliefs developed during an intervention are important for maintenance of an exercise regimen.