Tibial osteotomy for correction of angular deformity associated with arthrosis was performed in sixty-three knees in fifty-eight patients reviewed one to three years after operation. Half of the patients were older than sixty-five years with a range of forty through eighty-four years. Before and after operation the femorotibial angle was determined by roentgenographic examination in weight-bearing, standing position, and objective evidence of instability of the knee in the frontal plane was recorded. At operation a wedge of cancellous bone was removed proximal to the tibial tuberosity. The patients walked in a plaster cylinder during the immediate postoperative period and bending exercises were started at six weeks. Before the operation the femorotibial angle was more than 177 degrees in the fifty-three varus knees and less than 164 degrees in the ten valgus knees. After the operation forty of the sixty-three knees had a femorotibial angle of 164 to 177 degrees and twenty-three knees were outside these limits due to undercorrection or overcorrection of the deformity. In the former group all were stable, in the latter, only five. Ten of the eighteen unstable knees were painful in contrast to only three of the forty-five stable knees. These observations are interpreted to mean that pain in osteo-arthritis of the knee is due to instability, increasing with degree of deformity. Indications for corrective osteotomy in gonarthrosis should, therefore, include not only gross deformity, but also barely perceptible instability. The rationale for this approach is supported by cineroentgenography and radionuclide scintimetry. Advanced age, cardiopathy, or gross obesity did not seem to constitute contraindications.