The cases of twenty-one consecutive patients who had a minimally constrained total knee arthroplasty (six of whom had a cemented and fifteen, an uncemented prosthesis) after a failed proximal tibial osteotomy for osteoarthritis were compared with those of a non-consecutive group of twenty-one patients who had had a primary total knee arthroplasty for osteoarthritis. The groups were matched according to age and sex of the patient, type of prosthesis and fixation, and length of follow-up. At an average length of follow-up of 2.9 years, a good or excellent result was obtained in 81 per cent of the patients who had had a previous osteotomy. At an average length of follow-up of 2.8 years, a good or excellent result was obtained in 100 per cent of the patients who had had a primary arthroplasty. Two patients in the osteotomy group and none in the primary arthroplasty group required additional surgery. At the time of arthroplasty, technical difficulties in exposing the proximal part of the tibia were noted in three patients in the group that had undergone an osteotomy. The results of total knee arthroplasty after failed proximal tibial osteotomy approached but did not equal the results after primary total knee arthroplasty.