Despite the efficacy of a range of behavioral medicine interventions, high rates of attrition are a persistent problem in both clinical and research settings. Appropriately, studies have begun to focus on predictors of attrition with the hope of identifying important client or treatment characteristics. This article reviews attrition predictors in outpatient behavioral medicine treatments for headache, pain, stress, and weight management. Across all areas, psychological variables and severity of symptom variables were more predictive than demographic variables. However, as 13 of the 20 studies reviewed were in the area of weight management, generalizability of the findings to other treatment areas requires further investigation. Recommendations are made for improving attrition research by (a) developing clinically valid definitions of attrition, (b) recognizing important within-group differences among those who prematurely terminate treatment, and (c) focusing on theoretically grounded psychological and treatment process variables. A working definition of attrition based on the integration of clients' and clinicians' perspectives is also provided.