Traumatic anterior shoulder dislocation and sub-luxation are common injuries. But few studies have compared arthroscopic and open stabilization of the shoulder at long-term follow up. The purpose of our study is to show whether an arthroscopic approach to repair Bankart lesion can obtain the same results at long follow up as an open procedure. We analyzed 110 non-randomized consecutive shoulders in 110 patients who underwent a surgical repair of recurrent anterior shoulder instability between 1990 and 1999. Eighty-two patients were available at long term follow up (74,5% retrieval rate). In particular, 49 patients (59.8%) (group A) were treated with arthroscopic transglenoid suture (modified Caspari) between 1990 and 1995 (mean 15,7 year FU), whereas, 33 patients (40.2%) (group B) were treated with open repair between 1995 and 1999 (mean 12,7 year FU). We evaluated the patients in terms of failure rates, Rowe and UCLA scores. The failure cases in the forty-nine patients treated with arthroscopic suture were 13, six dislocations and seven subluxations. The group A had also a Rowe score: function 24.2+8.2, stability 42.4+13.9, range of movement 18.6+3.8, total score 85.0+22.46. The UCLA score was: pain 8.8+1.7, function 8.6+2.1, muscle power 9.2+1.6, total score 26.4+4.8. Of the thirty-three patients treated with open repair, three had at least one post-op dislocations and four felt sometimes subluxations. The Rowe score in group B was: function 23.6+9.7, stability 41.2+14.9, range of movement 18.3+3.9, total score 83.2+24.4. Moreover the UCLA score was: pain 8.8+1.9, function 8.8+1.9, muscle power 9.2+1.2, total score 26.9+4.2. We showed that both techniques were fairly good in treatment of shoulder instability. In our series no significant difference was observed in redislocation rate and in Rowe and UCLA scores between the two groups. The recurrence rate (subluxations and dislocations) was high in both groups: the arthroscopic group had 26.5% and the open one had 21.2%. Our recurrence rate following open repair was higher than in many studies, while the rate after arthroscopic transglenoid procedure was almost equivalent. We hypothesize that one of the reasons for these higher recurrence rates may be the long term follow up. Another cause could be our decision to include subluxation as a failure value, even if there is no agreement about. In fact we believe it to be an important disability factor in sport as in life activities. After surgery, most of the patients returned to their preinjuried activities. But at long term follow up almost all patients have stopped high level sport activity. Moreover, at this long term follow up, some patients told us a feeling of muscle weakness in the last years. In conclusion patients had good impressions about their shoulders thanks to surgery, but also because of lower functional demand.