We reviewed thirty-two arthrodeses of the ankle in twenty-six patients who had rheumatoid arthritis. In seventeen patients (eighteen ankles), a compression arthrodesis was done and external fixation was used. In eight patients (twelve ankles), we used internal fixation with 6.5-millimeter cancellous-bone screws. In the remaining patient, an arthrodesis with external fixation was done in one ankle and internal fixation was used in the other ankle; data for the appropriate ankle are included in each group. The patients were followed for an average of thirty-three months. The two groups were comparable with respect to age, sex, preoperative medications, and severity of disease. The average time to fusion was nineteen weeks in the compression arthrodesis group and seventeen weeks in the internal fixation group. Of the nineteen ankles that had a compression arthrodesis, four failed to fuse; all of the failures were associated with infection. Infection developed in two additional patients, there was malposition of the fusion in three patients, and neurapraxia developed in three patients. Of the thirteen ankles that had internal fixation, three ankles failed to fuse; one of the failures was associated with infection. Infection developed in one additional ankle. In two patients, the ankle fused in excessive valgus. Comparison of the two groups revealed comparable rates of fusion: fusion occurred in fifteen of the nineteen ankles in the group that had compression arthrodesis and in ten of the thirteen ankles in the group that had internal fixation. The method of arthrodesis did not affect the time to fusion or the rate of complications.(ABSTRACT TRUNCATED AT 250 WORDS)