We hypothesized that a pattern of decreasing postoperative length of stay (LOS) in the hospital could lead to an increased rate of manipulation after total knee arthroplasty (TKA). The decision to manipulate is based on the patient's ability to perform normal physical activities, which may in large part be dependent on adequate knee flexion (ie, range of motion [ROM]). Decreased exposure to physical therapy (as a result of decreased LOS) may be a contributing factor leading to impaired functional ROM in the 6-week postoperative period. We examined records from 745 patients (2 surgeons) who had a primary TKA from 1993 to 1996. At our institution, development and implementation of clinical pathways resulted in a significant decrease in the average LOS beginning in 1993. The average LOS in 1993 was 6.4 +/- 1.8 days (mean +/- SD) and progressively decreased to 4.4 +/- 1.0 days in 1996 (P < .0001). The rate of manipulation (patients manipulated at 6 weeks/total number of patients receiving TKA) was 6.0% in 1993 and increased to 11.3% (P = .09) in 1994, 13.5% (P = .02) in 1995, and 12.0% (P = .05) in 1996. In the period 1993-1996, patients requiring manipulation consistently had a lower ROM at discharge from the hospital (69.0 +/- 10 degrees, n = 67, P < .0001) compared with patients not requiring manipulation (80.7 +/- 10.6 degrees, n = 542). The relatively low frequency and charge for manipulation may justify the decision to decrease patient exposure to physical therapy through reduction in LOS. As efforts to decrease LOS continue, however, we must aggressively manage patients and provide adequate exposure to inpatient and outpatient physical therapy to ensure optimal ROM results.