Few studies have examined diabetes impact on total joint arthroplasty (TJA) outcomes, with variable findings. We investigated the association between diabetes and post-TJA physical function and pain, examining whether diabetes impact differs by sex and BMI. Patient sample completed questionnaires within 3 months prior to hip or knee TJA for osteoarthritis (OA) and 1-year post-surgery. Surgical ‘non-response’ was defined as < 30% improvement in Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain and physical function at 1-year. Two adjusted logistic regression models were estimated: (1) excluding, (2) including an interaction between diabetes, sex and BMI. The sample (626 hip, 754 knee) was 54.9% female, had mean BMI of 30.1, 13.0% reported diabetes. In adjusted models excluding an interaction, diabetes was not associated with non-response. However, a significant 3-way interaction (physical function: p = 0.003; pain: p = 0.006) between diabetes, sex, and BMI was found and was associated with non-response: non-response probability increased with increasing BMI in men with diabetes, but decreased with increasing BMI in women in diabetes. Findings suggest uncertainty in diabetes impact may be due to differential impacts by sex and BMI. A simple consideration of diabetes as present vs. absent may not be sufficient, with implications for the large TJA population.