Although arthroscopic Bankart repair has become an accepted surgical stabilization technique for anterior shoulder instability, the failure rate remains unacceptably high. Little information is available concerning healing of the Bankart repair. The purpose of this article is to clarify this issue by analyzing a cohort of 15 patients who underwent a "second-look" arthroscopy to evaluate and treat pain or recurrent instability following arthroscopic Bankart repair with the Suretac device (Acufex Microsurgical, Mansfield, MA). "Second-look" arthroscopy was performed at an average of 9 months following the index surgical procedure. The reasons for this second surgery were recurrent instability in 7, pain in 6, and pain and stiffness in 2. In the 7 patients with recurrent instability, the Bankart repair was found to be completely healed in 3 (43%), partially healed in 1 (14%), and had recurred in 3 (43%); however, 6 of 7 were observed to have lax capsular tissue. In 4 of these cases, retrospective review of the index surgical procedure showed that a technical error had been made during the repair. Two cases had biopsy of the repair site on "second-look" at 6 to 8 months, and this showed residual polyglyconate polymer debris surrounded by a histiocytic infiltrate. In the remaining 8 cases with stable shoulders, the Bankart repair had completely healed in 5 cases (62.5%) and partially healed in 3 cases (37.5%). The higher failure rate with this approach compared with open approaches appears to result from improper patient selection and errors in surgical technique. There is some question concerning healing strength of the Bankart repair, although complete healing of the Bankart does not seem to be a prerequesite for shoulder stability. Success of the procedure might be expected to improve by selecting only patients with unidirectional, posttraumatic, anterior instability who are found to have a discrete Bankart lesion and well-developed ligamentous tissue.