We report a prospective randomly controlled trial to examine the effectiveness of continuous passive motion (CPM) in improving postoperative function and range of movement after total knee arthroplasty (TKA). We allocated 53 patients (57 knees) to one of three postoperative regimes: no CPM (n = 19); CPM at 0 to 40 degrees (0 to 40 CPM; n = 18); and CPM at 0 to 70 degrees (0 to 70 CPM; n = 20). Those in the CPM groups had CPM for 48 hours and all patients had an identical regime of physiotherapy. There was an even distribution of various cemented and cementless TKAs in each group. Patients were assessed preoperatively and at one week and at one year postoperatively. At one week, there was a statistically significant increase in the range of flexion and total range of movement in the 0 to 70 CPM group compared with the no-CPM group. At one year we found no significant differences in mean flexion, overall range of movement, fixed flexion deformity or functional results in the three groups. Those who had CPM had a significant increase in analgesic requirement (p = 0.04). There was an increased mean blood drainage postoperatively in those who had 0 to 70 CPM (1558 ml) compared with those with no CPM (956 ml) (t = 2.96, p = 0.005) and with 0 to 40 CPM (1017 ml) (t = 2.62, p = 0.01). Our findings show that CPM had no significant advantage in terms of improving function or range of movement, and that its use increased blood loss and analgesic requirements.