The purpose of this study was as follows: 1) to review our previous findings regarding surgical outcome following open retinacular release but now at a mean followup of 4 years, 2) to test the validity of the correction of patellar "overhang" by Merchant views at 4 years and correlate this with outcome, 3) to correlate arthroscopic findings of lateral overhang, patellofemoral cartilage disease, preoperative Q-angle measurements, and duration of symptoms with outcome at 4 years postrelease. The surgical technique of diagnostic arthroscopy and open longitudinal release has been described previously by the author, emphasizing the specific anatomical structures to be divided. The synovium was never incised and the retinacular structures were freed to allow reduction of the patella medially. Sixty patients comprising 76 knees were reviewed at a mean followup of 4 years postrelease. Followup included evaluation of Merchant views and Q angles, physical examination, and determination of patient satisfaction. Eighty-five percent of the patients followed retained the good to excellent ratings they had at the 2 year followup. "Reduction" of the released patella on Merchant views correlated well with surgical outcome and patient satisfaction in 95% of cases. Residual Merchant view patellar tilt or excessive lateral position correlated well with poor results. Hematoma formation was the leading cause of failure. A severe preoperative Q angle correlated somewhat with results at 4 years but this was not statistically significant. The degree of articular cartilage disease at prerelease arthroscopy did not correlate at all with results at 4 years, nor did the duration of preoperative symptoms.(ABSTRACT TRUNCATED AT 250 WORDS)