During the five years from 1973 through 1977, eighty Mayo and Coonrad total elbow arthroplasties were performed in seventy-two patients at the Mayo Clinic. Follow-up after at least two years (average, four years) revealed that the results were good in 60 per cent, fair in 16 per cent, and poor in 24 per cent. Pain was a major symptom in 80 per cent of the elbows preoperatively but in only 3 per cent postoperatively. At follow-up, motion had increased: extension-flexion by 10 degrees (average range, 29 to 131 degrees of flexion) and forearm rotation by 26 degrees (average range, 61 degrees of pronation to 59 degrees of supination). Excluding eleven prostheses with loosening for which revision was necessary, there were forty-four complications (55 per cent) after eighty procedures. Of these forty-four complications, eleven were ulnar neuropathies (two permanent and nine transient); four were wound-healing problems; ten, significant triceps weakness; eleven, intraoperative fractures of the medial or lateral supracondylar bone column; seven, deep infections; and one was an ulnar fracture. Although the complication rate was very high, most of the complications occurred during the early years of the study. Follow-up revealed twenty-nine elbows with radiolucency around the components: in twenty-five about the humeral and in four about the ulnar component. Revisions were performed in nineteen (24 per cent) of the eighty elbows: in eleven because of loosening, in seven because of deep infection, and in one because of ankylosis. Intraoperative supracondylar fracture and defective cementing of the prosthesis were important factors contributing to prosthetic loosening. When the arthroplasty was successful, the relief of pain was dramatic, stability was excellent, and the range of motion was superior to that provided by any other procedure currently available.