Objective: To examine the association between obesity and knee and hip arthroplasty for osteoarthritis across a range of physical performance. Method: The body mass index and physical performance (on the 36-item Short Form Health Survey) of 9135 Australian Diabetes, Obesity and Lifestyle Study participants were measured in 1999–2000. The incidence of knee and hip arthroplasty during 2002–2011 was determined by linking the cohort records to the Australian Orthopaedic Association National Joint Replacement Registry. Results: Over 9.1 ± 2.3 years (mean ± sd)) of follow-up, 317 participants had knee and 202 had hip arthroplasty for osteoarthritis. Using those with neither obesity nor significantly impaired physical performance as the reference group, participants with both obesity and significantly impaired physical performance had a higher knee arthroplasty risk [hazard ratio (HR) = 5.25, 95% confidence interval (CI) 3.85–7.14] than those with obesity alone (HR = 2.49, 95% CI 1.81–3.44) or impaired physical performance alone (HR = 2.19, 95% CI 1.59–3.02). Similar results were observed for hip arthroplasty (obesity and impaired physical performance: HR = 2.67, 95% CI 1.72–4.15; obesity alone: HR = 1.65, 95% CI 1.08–2.51; impaired physical performance alone: HR = 1.83, 95% CI 1.26–2.66). Among overweight/obese patients, 5 kg greater baseline weight increased the knee arthroplasty risk across all levels of physical performance, and hip arthroplasty risk in those with the highest level of physical performance. Conclusion: Although impaired physical performance is an independent risk factor for knee and hip arthroplasty, greater weight increased knee arthroplasty for overweight/obese participants at all levels of physical performance, but hip arthroplasty only in those with good physical performance. Targeting weight loss has the potential to reduce the risk of knee arthroplasty and improve patient outcomes, even in those with poor physical performance.