We analyzed the results of seventy-one arthroscopies of the elbow in seventy patients who had been followed for an average of thirty-four months, in order to evaluate the risks and benefits of the procedure. Thirty-four arthroscopies were done for diagnostic purposes, fifteen were done for treatment, and twenty-two were done for both diagnosis and treatment. Overall, fifty-one (73 per cent) of the seventy patients benefited in some way. There were diagnostic benefits in thirty-six (64 per cent) of the fifty-six elbows in which diagnosis was an indication for the procedure, and there were therapeutic benefits in thirty (70 per cent) of the forty-three elbows in which treatment was intended or was performed although not planned. The procedure was of benefit in only eighteen (75 per cent) of the twenty-four elbows that had loose bodies, but it was successful in all elbows in which the loose bodies were not secondary to some other condition, including arthrosis. The procedure was also successful in all four elbows in which the loose bodies had been due to osteochondritis dissecans. In twelve (80 per cent) of the fifteen patients who had débridement (removal of flaps or loose fragments of articular cartilage) and in one of the two in whom a synovectomy had been performed, the treatment was successful. Seven (10 per cent) of the seventy patients had complications, none of which were major. Three patients (4 per cent) had a transient radial-nerve palsy after intra-articular injection of local anesthetic; four others had persistent drainage and negative cultures, but the drainage resolved with antibiotic therapy. In one of the four patients, a permanent flexion contracture of 15 degrees developed, and 10 degrees of flexion was lost.