INTRODUCTION: Limited quantitative information exists about the patient and surgeon factors driving variation in patient-reported outcome measure (PROM) scores, limiting the use of these data in understanding and improving quality. The overall goal of this study was to learn how to adjust PROM scores to enable both individual and group quality improvement. METHODS: Observational study in which preoperative Oxford Knee Score (OKS) and Patient Reported Outcomes Measurement System (PROMIS)-10 measures were prospectively obtained through patient survey from 1,173 of 1,435 possible patients before total knee arthroplasty and from 810 of the 1,173 patients at 12 months postoperatively (response rates = 81.7% and 69.0%). Regression analyses identified the relative contribution of patient and surgeon risk factors to OKS change from baseline to 12 months. Variation in patient scores and surgeon performance was described and quantified. Adjusted outcomes were used to calculate an observed and expected score for each surgeon. RESULTS: (1) Moderate variation was observed in pre-/post-OKS change among the surgeons (n = 16, mean change = 15.5 ± 2.2, range = 12.1-21.1). Forty-five percent of the variance in OKS change was explained by the factors included in our model. (2) Patient preoperative OKS and PROMIS physical score, race, and BMI were markedly associated with change in OKS, but other patient factors, surgeon volume, and years of experience were not. (3) Eight surgeons had observed scores greater than expected after adjustment, providing an opportunity to learn what strategies were associated with better outcomes. DISCUSSION: Traditional age/sex adjustment of patient mix would have had no effect on mean PROM scores by surgeon. An adjustment model that includes the factors found to be markedly associated with outcomes will allow care systems to identify individual surgeon care management strategies potentially important for improving patient outcomes.