We reviewed 100 patients treated arthroscopically for symptoms of chronic ankle pain associated with sprains of the ankle. All had pain that had failed to respond to conservative treatment for at least 6 months. The pathology in 95 of the 100 ankles studied could be categorized into one of three groups: the instabilities (lateral and syndesmotic), the impingements (anterior and anterolateral), and articular lesions (chondral and osteochondral). Five patients had nonspecific osteoarthritis and/or synovitis on arthroscopy. Patients were followed-up for improvements in six categories: pain, swelling, stiffness, limping, activity, and instability. The primary outcomes of pain and activity were analyzed statistically. Patient satisfaction and return to sports were evaluated. Significant improvements were obtained for patients treated for syndesmotic instability, and anterior and anterolateral impingement. Chondral fractures in the presence of a stable ankle had good results in 75% of cases, compared with those in unstable ankles with only 33% good results. Osteochondritis dissecans was treated successfully by excision of the lesion and abrasion of the base. Patients with chronic lateral instability were treated by open repair, so only the diagnostic arthroscopic findings are reported. We concluded that arthroscopy offered little to the management of lateral instability unless there was considerable doubt regarding the diagnosis. There were minimal improvements for the patients with nonspecific diagnoses such as posttraumatic synovitis. Ankle arthroscopy may be a very useful diagnostic and therapeutic tool in patients who have not responded to conservative therapy.