Clinic records of 37 manipulations in 767 consecutive primary total knee arthroplasties (TKAs) were analyzed to identify any predictors of manipulation outcome. Factors studied were sex, age, body mass index, tibiofemoral alignment, surgical history, smoking history, range of motion before TKA and manipulation, intraoperative lateral release, implant design and manufacturer, and manipulation interval. Measures of outcome were gains in extension and flexion from manipulation and range of motion at 1-year follow-up. Patients gained an average of 4 degrees of extension and 22 degrees of flexion after manipulation, resulting in average extension of 1 degree and average flexion of 105 degrees at 1-year follow-up. Restored flexion was similar to that measured preoperatively. Manipulation was most effective in patients manipulated within 8 weeks, with full extension and <90 degrees of flexion prior to manipulation, and those receiving a lateral release during arthroplasty. Potential for benefit from manipulation appears to be reduced in patients with large flexion contractures but with adequate flexion. Patients with flexion contractures regained only approximately 80% of the extension they lacked before manipulation. Only 12 of 18 patients (67%) with flexion contractures regained full extension. Manipulation is successful at regaining flexion and restoring function for patients of all levels of flexion deficiency. At least 90 degrees of flexion was regained in 20 of 23 patients (87%) who lacked it at manipulation.