To investigate whether the performance of the short form-12 (SF-12) health survey is comparable with the longer version SF-36 for measuring health-related quality of life over time in patients with knee osteoarthritis who have joint replacement surgery.Four hundred and seven patients with knee osteoarthritis completed the SF-36 before surgery and at a minimum of 12 months following knee replacement. SF-12 item responses were obtained from the responses given to the SF-36 questionnaire. Correlation coefficients were calculated between SF-12 and SF-36 physical component summary (PCS) and mental component summary (MCS) scores and the respective change in scores. Sensitivity to change was determined with the standardised response mean (SRM).PCS and MCS scores were highly correlated between SF-12 and SF-36 versions for both preoperative and post-operative measures (r = 0.90-0.96, p < 0.0001). Change scores (post-operative-preoperative) were also highly correlated (PCS: r = 0.88, p < 0.001;r = 0.93, p < 0.001). Sensitivity to change was large for the PCS scale (all SRMs >1.0). Correlations above 0.7 were found between change scores for each SF-36 and SF-12 subscale except General Health (r = 0.55).The SF-12 summary measures and component scores replicate well with the SF-36 and show similar responsiveness to change. The SF-12 appears to be an adequate alternative for use in patients with knee osteoarthritis who undergo replacement surgery, and its brevity should be attractive for both clinicians and patients.I.