We compared the results of twenty-three consecutive capitellocondylar total elbow arthroplasties in twenty-three patients in whom an excision of the radial head and synovectomy for rheumatoid arthritis had failed with those of twenty-three non-consecutive primary capitellocondylar total elbow arthroplasties in twenty-three patients who had rheumatoid arthritis. The two groups were matched for age, gender, duration of follow-up, side of the operation, type of prosthesis, and operative approach. The average duration of follow-up was four years (range, two to fourteen years). At the most recent follow-up examination, use of a 100-point rating system demonstrated an improvement from an average preoperative score of 21 points (range, 12 to 42 points) to an average postoperative score of 87 points (range, 17 to 97 points) for the group in whom an excision of the radial head and synovectomy had failed. The group that had primary arthroplasty demonstrated an improvement from an average preoperative score of 22 points (range, 7 to 42 points) to an average postoperative score of 94 points (range, 85 to 100 points). The group that had primary arthroplasty had a significantly greater improvement in terms of relief of pain (p < 0.05), functional status (p < 0.01), and the elbow-rating score (p < 0.03) than the other group. Four patients who had had failure of an excision of the radial head and synovectomy and none of those who had primary arthroplasty needed an additional operative procedure. Six of the patients who had had a failed excision and synovectomy and none of the patients who had primary arthroplasty had instability of the elbow components. We concluded that, although excision of the radial head and synovectomy is a conservative and effective method of treating a painful rheumatoid elbow, conversion to a capitellocondylar total elbow arthroplasty is more difficult after such an operation and the results at a minimum of two years are inferior to those for primary capitellocondylar total elbow arthroplasty.