The biomechanical response and injury tolerance of the shoulder in lateral impacts is not well understood. These data are needed to better understand human injury tolerance, validate finite element models and develop biofidelic shoulders in side impact dummies. Seventeen side impact sled tests were performed with unembalmed human cadavers. Data analyzed for this study include T1-Y acceleration, shoulder and thoracic load plate forces, upper sternum x and y accelerations, and struck side acromion x, y and z accelerations. One dimensional deflection at the shoulder level was determined from high-speed film by measuring the distance between a target on T1 and the impacted wall. Force-time response corridors were obtained for tests with 9 m/s pelvic offset, 10.5 m/s pelvic offset, 9 m/s unpadded flat wall, 6.7 m/s unpadded flat wall, 9 m/s soft padding and 9 m/s stiff padding. Maximum shoulder plate forces in unpadded 9 m/s tests (5.5 kN) were larger than in 6.7 m/s tests (3.3 kN). The peak force at the shoulder was larger than at the thorax plate in unpadded and soft padded tests. T1-Y accelerations were larger in unpadded 9 m/s flat wall tests and unpadded pelvic offset 10.5 m/s tests (peak values of 130 and 145 g’s) than in other test conditions. Deflections between T1 and the struck wall ranged from 88 to 154 mm in unpadded tests and 95 to 128 mm in stiff and soft padded tests. Eighteen AIS 2 level shoulder injuries occurred in 11 test subjects. These injuries included left acromion fracture in five subjects, left acromioclavicular separation in ten subjects and left clavicle fracture in three subjects. Average MAIS to the shoulder was 0.86 in seven subjects which impacted 4 to 6 inches (101.6 to 152.4 mm) of soft or stiff padding and 1.6 in ten subjects which impacted no padding or 3 inches (76.2 mm) of stiff padding. Previous findings from this test series were reported by Irwin et al. (1993) for seven tests and focused on detailed analysis of shoulder deflection (T5 to shoulder edge). The current study is expanded to 17 tests and includes force, acceleration response and analysis of shoulder deflection (T1 to shoulder edge). Padding of 4 to 6 inches (101.6 mm to 152.4 mm) reduced shoulder injury approximately one AIS level. A combination of ASA10 and deflection was the best shoulder injury predictor. Shoulder deflection of 106 mm predicts 50% probability of MAIS 2.