The authors compare the clinical and radiographic outcome in patients with comparable bilateral recurrent patellar dislocation treated surgically on only one side, to clarify the appropriateness of the surgical indication. Sixteen patients were evaluated at an average follow-up of 30 years (20-45); all had been treated by the Roux technique. The results on both the operated and the unoperated knee were evaluated; the Crosby and Insall schedule was used for the clinical evaluation. Anteroposterior, lateral, and Merchant's view X-rays were examined for osteoarthritis and to measure the height of the patella. The congruence angle and the distance between anterior tibial tuberosity and trochlear groove (ATT-TG) were measured by computed tomography. The results in the operated knees were: 3 excellent, 9 good, 1 fair, and 3 worse; results in the nonoperated knees were 6 excellent, 8 good, 1 fair, and 1 worse. In the operated knees arthritis was grossly marked in 8, marked in 3, moderate in 1, and light in 4; in the nonoperated ones it was grossly marked in 8, moderate in 3, and light in 5. The congruence angle was normal in 10, medially displaced in 3, and laterally displaced in 3 cases on the operated side; on the nonoperated side it was normal in 7 cases and lateralized in the remaining 9. The ATT-TG in the operated knees was negative in 9 cases, normal in 1, and positive in 6; on the non-operated side it was positive or normal. In 7 operated cases a low patella was documented. The Roux technique yields positive results in the correction of recurrent dislocation. No clinical or radiographic differences were found between surgically and conservatively treated knees. The clinical results are generally not comparable with the radiographic features or with severity of degenerative modifications presented at long-term follow-up. The absence of a difference is due basically to the complete lack of adaptation of the surgical procedure to the variable pathogenesis of this disorder.