Approximately 50% of people with eating disorders (EDs) engage in driven exercise to influence their weight or shape and/or to compensate for loss-of-control eating. When present, driven exercise is associated with a lower quality-of-life, longer hospital stays, and faster rates-of-relapse. Despite the seriousness of driven exercise, most treatments for EDs do not target maladaptive exercise behaviors directly. Given the large proportion of patients with an ED who engage in driven exercise and its effect on treatment outcomes, it is critical to understand what predicts change in driven exercise. The purpose of this study was to test whether ED symptoms prospectively predicted change in driven exercise and vice versa.