Background: An initial anterior dislocation of the shoulder becomes recurrent in 66% to 94% of young patients after immobilization of the shoulder in internal rotation. Magnetic resonance imaging and studies of cadavera have shown that coaptation of the Bankart lesion is better with the arm in external rotation than it is with the arm in internal rotation. Our aim was to determine the benefit of immobilization in external rotation in a randomized controlled trial. Methods: One hundred and ninety-eight patients with an initial anterior dislocation of the shoulder were randomly assigned to be treated with immobilization in either internal rotation (ninety-four shoulders) or external rotation (104 shoulders) for three weeks. The primary outcome measure was a recurrent dislocation or subluxation. The minimum follow-up period was two years. Results: The follow-up rate was seventy-four (79%) of ninety-four in the internal rotation group and eighty-five (82%) of 104 in the external rotation group. The compliance rate was thirty-nine (53%) of seventy-four in the internal rotation group and sixty-one (72%) of eighty-five in the external rotation group (p = 0.013). The intention-to-treat analysis revealed that the recurrence rate in the external rotation group (twenty-two of eighty-five; 26%) was significantly lower than that in the internal rotation group (thirty-one of seventy-four; 42%) (p = 0.033) with a relative risk reduction of 38.2%. In the subgroup of patients who were thirty years of age or younger, the relative risk reduction was 46.1%. Conclusions: Immobilization in external rotation after an initial shoulder dislocation reduces the risk of recurrence compared with that associated with the conventional method of immobilization in internal rotation. This treatment method appears to be particularly beneficial for patients who are thirty years of age or younger. Level of Evidence: Therapeutic Level II. See Instructions to Authors for a complete description of levels of evidence.