Background. The functional outcome and risk of recurrence following arthroscopic stabilisation for recurrent anterior shoulder instability is poorly defined in large prospective outcome studies. This is the first study to prospectively evaluate these outcomes in patients who have been treated using this technique. Methods. We performed a prospective study of a consecutive series of 302 patients (265 men and 37 women, mean age 26.4 years) who underwent 311 (9 bilateral) arthroscopic Bankart repairs for recurrent anterior instability. Patients were evaluated preoperatively and postoperatively at 6 months, and annually thereafter. The chief outcome measures were risk of recurrence and the two-year functional outcomes (assessed using the WOSI and DASH scores). Results. On survival analysis, the overall re-dislocation rate after surgery was 13.5% (42/311 shoulders). The median time to recurrence was 12 months (range 3 to 110 months) and 55% of these developed recurrent instability within 1 year of their surgery. The median follow-up in those patients who did not re-dislocate was 70 months. There was a significant improvement in the WOSI and DASH scores at 2 years postoperatively of 21.4 and 9.9 points respectively (both p<0.001). The risk of redislocation was significantly increased by the presence of glenoid bone loss associated with or without a Hill-Sachs lesion, and the type and engagement of a Hill-Sachs lesion on arthroscopic examination (all p<0.001). However, the risk in these sub-groups was lower than that reported in previous studies. A multivariate model was produced from these parameters to estimate the risk of recurrence at the time of primary surgery. Conclusions. The majority of patients have successful functional outcomes and relatively low risk of recurrence following arthroscopic stabilisation. The identification of preoperative factors which increase the risk of instability has enabled us to better counsel patients of their risk of failure following arthroscopic repair