Past experience has shown that the results of high tibial osteotomy are difficult to predict. The purpose of this review was to correlate preoperative findings with postoperative results to define more precisely the indications and contraindications and to improve results. Between 1970 and 1983, 72 high tibial osteotomies for osteoarthritis were done at Duke University Medical Center by one surgeon using a consistent surgical technique. Fourteen complications were identified in 72 knees. The only one causing a persistent problem was a fracture of the tibial plateau. Forty-five knees in 38 patients were available for follow-up at an average of six years (range 1.5 to 15.0 years). Results in 15 knees (33%) were classified as "good" (pain free), in 20 knees (44%) as "fair" (improved), and in ten knees (22%) as "poor" (unchanged or worse). Three patients with fair results subsequently had total knee arthroplasty at least nine years after the osteotomy, and two of those with poor results received total knee replacement within two years of the tibial osteotomy. Measurements obtained from standing roentgenograms revealed an average of 3 degrees of preoperative varus and 8 degrees of postoperative valgus, for an average correction of 11 degrees. There was no statistically significant correlation between preoperative measurements of joint space or alignment and postoperative result. This review indicates that preoperative standing films of the knee alone are not a reliable indication for patient selection.