In far-side impacts, head contact with interior components is a key injury mechanism. Restraint characteristics have a pronounced influence on head motion and injury risk. This study performed a parametric examination of restraint, positioning, and collision factors affecting shoulder belt retention and occupant kinematics in far-side lateral and oblique sled tests with post mortem human subjects (PMHS). Seven PMHS were subjected to repeated tests varying the D-ring position, arm position, pelvis restraint, pre-tensioning, and impact severity. Each PMHS was subjected to four low-severity tests (6.6 g sled acceleration pulse) in which the restraint or position parameters were varied and then a single higher-severity test (14 g) with a chosen restraint configuration (total of 36 tests). Three PMHS were tested in a purely lateral (90° from frontal) impact direction; 4 were tested in an oblique impact (60° from frontal). All subjects were restrained by a 3-point seatbelt. Occupant motion was tracked with a 3D optoelectric high speed motion capture system. For all restraint configurations, the 60° oblique impact angle was associated with greater lateral head excursion than the 90° impact angle. This unexpected result reflects the increased axial rotation of the torso in the oblique impacts, which allowed the shoulder to displace more relative to the shoulder belt and thus the head to displace more relative to the sled buck. Restraint engagement of the torso and shoulder was actually greater in the purely lateral impacts than in the oblique impacts. Pretensioning significantly reduced lateral head excursion (175 mm average in the low-severity tests across all restraint configurations).