OBJECTIVES: Orthopedic researchers studying total joint arthroplasty (TJA) often use disease-specific measures (DSMs) of patient health as cost-effectiveness outcomes.  However, for cost-utility analysis, health utility scores are needed.  The study objective was to develop regression algorithms to map five common disease specific TJA outcome measures to three preference-based utility  scores.  METHODS: An online survey was completed by 438 total hip arthroplasty (THA) patients and 550 total knee arthroplasty (TKA) patients. THA patients completed the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC^®), Harris Hip Score (HHS), and the Hip Disability and Osteoarthritis Outcomes Score (HOOS). Knee patients completed the WOMAC^®, Knee Society Score (KSS), and Knee Disability and Osteoarthritis Outcomes Score (KOOS). All patients completed three preference based questionnaires, the SF-6D, EQ-5D and HUI-3, and responses were used to calculate health utilities. A total of 30 THA and 30 TKA mapping models were estimated for each possible pairing of utilities  and DSMs, and prediction performance was considered to pick the best DSM/utility match. RESULTS: For THA, the regression model with HOOS subscores most precisely estimated an EQ-5D utility. For the HOOS/EQ-5D model, the MAE was .05, the RMSE was .06 and the ME was nearly zero. The best performing TKA model mapped the KSS to the EQ-5D. The MAE was .05, the RMSE was .06 and the ME also approximated 0.00 for the KSS/EQ-5D. These prediction errors are small which indicates that these models can accurately estimate the EQ-5D utilities. However, they should only be used in conjunction with average patient characteristics and not with individual level data.  CONCLUSIONS: Clinicians and researchers can input their disease specific data into these models to estimate health utilities to consider the cost-effectiveness of osteoarthritis-related interventions relative to interventions for very different diseases and conditions.