Objective To examine whether the course of physical therapy treatments received by patients who undergo total knee replacement (TKR) surgery differs depending on whether the orthopedic surgeon has a financial stake in physical therapy services. Data Sample of Medicare beneficiaries who underwent TKR surgery during the years 2007–2009. Study Design We used regression analysis to evaluate the effect of physician self‐referral on the following outcomes: (1) time from discharge to first physical therapy visit; (2) episode length; (3) number of physical therapy visits per episode; (4) number of physical therapy service units per episode; and (5) number of physical therapy services per episode expressed in relative value units. Principal Findings TKR patients who underwent physical therapy treatment at a physician‐owned clinic received on average twice as many physical therapy visits (8.3 more) than patients whose TKR surgery was performed by a orthopedic surgeon who did not self‐refer physical therapy services (p < .001). Regression‐adjusted results show that TKR patients treated at physician‐owned clinics received almost nine fewer physical therapy service units during an episode compared with patients treated by nonself‐referring providers (p < .001). In relative value units, this difference was 4 (p < .001). In contrast, episodes where the orthopedic surgeon owner does not profit from physical therapy services rendered to the patient look virtually identical to episodes where the TKR surgery was performed by a surgeon nonowner. Conclusions Physical therapists not involved with physician‐owned clinics saw patients for fewer visits, but the composition of physical therapy services rendered during each visit included more individualized therapeutic exercises.