Objective: The incidence of joint replacements is considered an indicator of symptomatic end-stage osteoarthritis (OA). We analysed data from two national joint replacement registries in order to investigate whether evidence of a pattern of progression of end-stage hip and knee OA could be found in data from large unselected populations. Design: We obtained data on 78,634 hip and 122,096 knee arthroplasties from the Australian Orthopaedic Association National Joint Replacement Registry and 19,786 hip and 12,082 knee arthroplasties from the Norwegian Arthroplasty Register. A multi-state model was developed where individuals were followed from their first recorded hip or knee arthroplasty for OA to receiving subsequent hip and/or knee arthroplasties. We used this model to estimate relative hazard rates and probabilities for each registry separately. Results: The hazard rates of receiving subsequent arthroplasties in non-cognate joints were higher on the contralateral side than on the ipsilateral side to the index arthroplasty, especially if the index was a hip arthroplasty. After 5 years, the estimated probabilities of having received a knee contralateral to the index hip were more than 1.7 times the probabilities of having received a knee ipsilateral to the index hip. Conclusion: The results indicate that there is an association between the side of the first hip arthroplasty and side of subsequent knee arthroplasties. Further studies are needed to investigate whether increased risk of receiving an arthroplasty in the contralateral knee is related to having a hip arthroplasty and/or preoperative factors such as pain and altered gait associated with hip OA.