The use of an impulse-momentum (IM) exercise technique was investigated for end-stage shoulder rehabilitation. The objectives of this study were to: (a) quantify the net shoulder joint forces and moments while using an IM system and (b) test the influence of gender and muscle loading type (concentric or eccentric) on kinetic and kinematic parameters. Fourteen healthy adults (eight males, six females) performed a repeated measures experiment on an instrumented device utilizing a cabled shuttle system. While maintaining 90 degrees of shoulder abduction and 90 degrees of elbow flexion, the subjects externally rotated their upper arm from 0 degrees to 90 degrees (concentric acceleration) and then internally rotated their upper arm back from 90 degrees to the 0 degrees position (eccentric deceleration). Shoulder joint forces and moments as well as rotational work and power were calculated using inverse dynamics (free-body forces and moments calculated at intersegmental joint centres). Overall concentric peak forces and moments were greater than eccentric peak forces and moments (P < 0.0001). Joint forces and moments reached a maximum during the initial phase of concentric loading (0 degrees to 45 degrees) compared with any other rotational position in the loading cycle (concentric 45 degrees to 90 degrees or eccentric 90 degrees to 0 degrees). The results also indicate that males experienced higher (P < 0.0001) average resultant peak joint forces (concentric 0 degrees to 45 degrees = 108.0 N and eccentric 90 degrees to 45 degrees = 87.2 N) than females (concentric 0 degrees to 45 degrees = 74.7 N and eccentric 45 degrees to 0 degrees = 56.0 N). In addition, males experienced higher (P < 0.0001) average resultant peak joint moments (concentric 0 degrees to 45 degrees = 30.4 N m and eccentric 45 degrees to 0 degrees = 21.0 N m) than females (concentric 0 degrees to 45 degrees = 19.7 N m and eccentric 45 degrees to 0 degrees = 12.8 N m).