Objective: To describe the rates of participation in regular physical activity presurgery and at 3 years follow-up after knee or hip arthroplasty, and to describe factors associated with participation postsurgery and types of activity undertaken. Methods: A previously acquired multicenter, prospective cohort of knee or hip arthroplasty recipients was followed up for 3 years postsurgery. Regular participation in physical activity was defined as participation in physical activity ≥1 time/week, excluding incidental activities. Participants were interviewed about current participation as well as participation in the year presurgery. Joint-specific and health-related quality-of-life scores and information on experience of major complications were obtained. Information about comorbidity and body weight were updated. Factors associated with 3-year physical activity participation were determined using multivariable logistic regression modeling. Results: In total, 73.4% of the eligible cohort (1,289 of 1,757) were followed up (718 patients with total knee arthroplasty, and 571 patients with total hip arthroplasty). Participation profiles were similar regardless of the joint replaced. Participation in physical activity increased postsurgery in the combined cohort (from 45.2% to 63.5%; P < 0.001). Participation at 3 years was associated with participation presurgery (P < 0.0001), better 3-year quality of life (P < 0.001), younger age (P = 0.002), better 3-year joint scores (P = 0.01), >1 lifetime arthroplasty (P = 0.02), and higher education level (P = 0.04). Low-impact and nonambulatory activities significantly increased postsurgery with no change in high-impact activities. Conclusion: Participation rates increased postsurgery when recovery was stable, but approximately one-third of arthroplasty recipients did not engage in physical activity at least once per week. Because participation is associated with habitual activity presurgery, a potential role for behavior change interventions is suggested. The increase in nonambulatory activities indicates that current devices measuring ambulatory activity alone are inadequate for capturing physical activity.