Problem-solving skill is important for chronic illness self-management. This project prospectively evaluated a measure of diabetes problem-solving skill for its reliability, convergent validity, sensitivity to intervention, and relationship to change in behavior. Postmenopausal women with type 2 diabetes (N = 279) participated in a RCT to evaluate a lifestyle modification program. The 9-item Diabetes Problem-Solving Inventory (DPSI) was used to assess how patients cope with challenges to diabetes self-care. The DPSI was found to have good inter-rater reliability and internal consistency for a brief scale, be moderately stable over time, and relate significantly to hypothesized variables. DPSI scores improved significantly more in the lifestyle change condition than in controls and were related to improved outcomes. Mediation analyses indicated that the increase in problem-solving was a partial mediator of outcomes. Results support the reliability, predictive ability, and sensitivity to change of the DPSI. Directions for future research on problem-solving and chronic illness are discussed.