We report the week-to-week variation of the Oxford Knee and Oxford Hip Score (OKS, OHS) in individuals with severe osteoarthritis. People waitlisted for knee (n=51) or hip arthroplasty (n=52) were assessed twice, 1-week apart. There were no major week-to-week systematic biases in the scores. Limits of agreement (LOA) for both scores were wide (OKS, −9.5 to 6.6; OHS, −7.7 to 7). For most individual questions, the answers varied by ≤1 point in over 90% of participants. The week-to-week 95% LOA for the Oxford scores are unacceptably large, but variations within the individual questions are minimal. Consequently, reference to variation in the individual questions may be more useful for monitoring a patient's preoperative clinical change than changes in the total Oxford score. We conclude that the total scores are not suitable for monitoring the progression of OA in individual patients.