Purpose: The purpose of this study was to identify change in the exercise behavior and to identify factors influencing long-term adherence to aquatic exercise in patients with arthritis as a follow-up study after five years. Method: Baseline data had been collected using a structured interview 6 months after participants completed the aquatic exercise educational program. This second follow-up data collection included 133 participants from the original baseline study and was done 5 years after the original baseline study. Data was analyzed using SPSS 16 Win program. Results: The stage of change for the participants was as follows: Precontemplation (44.3%), Contemplation (24.1%), Preparation (3.0%), Action (0%) and Maintenance (adherence, 28.6%). The adherers (N=38) had significantly higher scores than non-adherers (N=91) in self-efficacy, outcome expectancy, group coherence and self-evaluation, and significantly lower scores in barriers to exercise. In logistic regression analysis, self-efficacy, group cohesion, barriers to exercise and self-evaluation were associated with exercise adherence (Chi Square=17.14, p=.002). Group cohesion (OR=2.5871, 95% CI=1.094-6.113, p= .030) was the most important factor. Conclusion: Self-efficacy, group cohesion, barriers to exercise and self-evaluation need to be considered in planning exercise programs to improve exercise adherence.