A goal-setting protocol, based on research in goal setting and performance and personal construct theory, was tested for its effect on adherence to a new exercise program. The Goal-setting group (n = 50) had significantly less dropout (30%) than the control group (n = 50) (74%). The Goal-setting group also had significantly better attendance (p<.0001). Suggestions for increasing confidence in findings through further research and practical implications of using the protocol to improve exercise maintenance across settings were discussed.