Objective: To assess self-reported physical activity changes pre- compared to post-operatively in patients undergoing total hip or knee arthroplasty, and to compare this to an age- and gender-matched cohort of people who have not undergone arthroplasty. Design: Population-based prospective cohort study. Setting: Norfolk, United Kingdom. Subjects: People who had undergone hip or knee arthroplasty, compared to an age- and gender-matched non-arthroplasty cohort. Intervention: Primary total hip or knee arthroplasty. Main measures: Physical activity, measured using the EPIC Physical Activity Questionnaire (EPAQ2). Results: A total of 400 people from the EPIC-Norfolk community cohort were identified who had undergone hip or knee arthroplasty. In all, 767 people were identified to form an age- and gender-matched non-arthroplasty cohort. Mean post-operative follow-up was 43 months post-total hip and 41 months post-total knee arthroplasty. There was a statistically significant reduction from pre- to post-arthroplasty in the number of flights of stairs climbed weekly (hip: mean difference (MD): 6.8; P < 0.01; knee: MD: 10.2; P < 0.01), duration of walking (hip: MD: 1.4 hours/week; P = .02; knee: MD: 2.2 hours/week; P < 0.01) and duration of total recreational activity (hip: MD: 1.1 hours/week; P = 0.02). Compared to the non-arthroplasty cohort, duration of physical activity was lower post-total hip arthroplasty (MD: 1.8 hours/week; P = 0.01). The number of flights of stairs climbed weekly (MD: 12.0; P < 0.01), total recreational activity (MD: 1.7 hours/week; P = 0.04) and physical activity energy expenditure (MD: 5.7 Mets-hours/week; P = 0.05) were lower for people post-total knee arthroplasty compared to the matched controls. Conclusions: Physical activity did not increase, and in some instances decreased, following total hip or knee arthroplasty.