Background Cutpoints (CPs) for mild, moderate and severe pain are established and used primarily in cancer pain. In this study, we wanted to determine the optimal CPs for mild, moderate, and severe pain in joint replacement surgery candidates with osteoarthritis (OA) of the hip or knee, and to validate the different CPs. Methods Patients (n = 353) completed the Brief Pain Inventory (BPI), the WOMAC Arthritis Index, and the SF-36 health status measure. Optimal CPs for categorizing average pain with three severity levels were derived using multivariate analysis of variance, using different CP sets for average pain as the independent variable and seven interference items from the BPI as the dependent variable. To validate the CPs, we assessed if patients in the three pain severity groups differed in pain as assessed with WOMAC and SF-36, and if BPI average pain with the optimal CPs resulted in higher correlation with pain dimensions of the WOMAC and SF-36 than other CPs. Results The optimal CPs on the 0–10 point BPI scale were CP (4,6) among hip patients and CP (4,7) among knee patients. The resulting pain severity groups differed in pain, as assessed with other scales than those used to derive the CPs. The optimal CPs had the highest association of average pain with WOMAC pain scores. Conclusion CPs for pain severity differed somewhat for patients with OA of the hip and knee. The association of BPI average pain scores categorized according to the optimal CPs with WOMAC pain scores supports the validity of the derived optimal CPs.