From 1993 through 1996, a multicenter study was conducted on the surgical treatment of patients with posttraumatic recurrent anterior shoulder dislocations. Fifty-six patients (40 men, 16 women; mean age 26 years [range 18-51 years]), were evaluated with shoulder arthroscopy. If a Bankart lesion was present, the patients were randomly allocated to either an arthroscopic reconstruction with the use of biodegradable tacks or an open reconstruction with suture anchors. The postoperative rehabilitation protocol for the two groups was identical. In all patients, the range of shoulder motion, stability, and the Constant and Rowe scores were evaluated at 3, 12, and 24 months postoperatively. Thirty patients were surgically treated with the arthroscopic technique and 26 patients with the open technique. In the arthroscopic group, there were recurrences in 7 (23%) of 30 patients at a mean of 13 months (range 5 to 21 months) after surgery. All patients with stable shoulders had a negative apprehension test result. In the open group, there were recurrences in 3 (12%) of 26 patients at a mean of 10 months (range 2 to 23 months) after surgery (P = not significant). In the arthroscopic group, 2 patients had new traumatic redislocations, whereas 1 patient redislocated during an epileptic seizure. In the open group, 1 traumatic redislocation occurred. The 2-year results in this study demonstrate a large number of redislocations after reconstruction, even in the open surgery group. Patient noncompliance with the rehabilitation protocol and predisposing disease may partially explain these results. A tendency was seen toward more redislocations in the arthroscopic group, which emphasizes the importance of correct patient selection and careful surgical technique in the difficult surgical procedure.