To date, several lateral impact studies ( Bolte et al., 2000 , 2003 , Marth, 2002 and Compigne et al., 2004 ) have been performed on the shoulder to determine the response characteristics and injury threshold of the shoulder complex. Our understanding of the biomechanical response and injury tolerance of the shoulder would be improved if the results of these tests were combined. From a larger data base shoulder injury tolerance criteria can be developed as well as corridors for side impact dummies. Data from the study by Marth (2002 , 12 tests) was combined with data from the previous studies. Twenty-two low speed tests (4.5 ± 0.7 m/s) and 9 high speed tests (6.7 ± 0.7 m/s) were selected from the combined data for developing corridors. Shoulder force, deflection and T1y acceleration corridors were developed using a minimization of cumulative variance technique. The reduction in impact speed and the addition of padding reduced the magnitude and increased the time to peak of shoulder forces and T1y accelerations. Logistic analyses were performed on the combined data sets to determine the best predictors of MAIS-2 shoulder injuries. Maximum normalized shoulder deflection and Cmax had p values of 0.0000 and were the best predictors of shoulder injuries. For the 50 th -percentile male, a shoulder deflection of 40 mm and a Cmax of 20% corresponded to a 50 % risk of MAIS-2 shoulder injury. In linear regression analysis, maximum normalized medial scapula × acceleration and maximum normalized sternum × acceleration were best related with the shoulder deflection and confirmed the forward movement of the sternum and rearward movement of the scapula.