Treatment with pulsing electromagnetic fields was used as an adjunct in twenty patients who had had a knee arthrodesis after failure of a total joint arthroplasty. Eighteen had had an infected arthroplasty; one, mechanical loosening; and one, recurrent dislocation. Arthrodesis had been attempted twenty-five times in these twenty patients prior to application of the coils. These procedures included the use of twenty-two external fixation frames, one compression plate, one intramedullary rod, and one cylinder cast. Two groups of patients were identified: those with non-union and those with delayed union. Fourteen patients began treatment six months or more after arthrodesis and were considered to have a non-union. The other six patients started treatment less than six months after attempted arthrodesis because there was no evidence of progression toward union. They were considered to have delayed union. In seventeen (85 per cent) of the twenty patients a clinically solid arthrodesis with roentgenographic evidence of bone-bridging was achieved. The average time to union after coil therapy was started was 5.8 months, with a range of three to twelve months. The patients who started coil treatment earlier after arthrodesis showed a tendency to heal faster. The three patients who had failures were the only ones who did not adhere to the protocol, and all three were in the non-union group. All patients with a solid arthrodesis were free of pain and able to walk at the time of follow-up, nine to thirty-one months after the completion of treatment. The use of pulsing electromagnetic fields appears to be a valuable non-invasive adjunct when performing arthrodesis of the knee after failed total joint arthroplasty.