Far-side side impact loading of a seat belt restrained occupant has been shown to lead to torso slip out of the shoulder belt. A pretensioned seat belt may provide an effective countermeasure to torso rollout; however the effectiveness may vary with age due to increased flexibility of the pediatric spine compared to adults. To explore this effect, low-speed lateral (90°) and oblique (60°) sled tests were conducted using male human volunteers (20 subjects: 9-14 years old, 10 subjects: 18-30 years old), in which the crash pulse safety envelope was defined from an amusement park bumper-car impact. Each subject was restrained by a lap and shoulder belt system equipped with an electromechanical motorized seat belt retractor (EMSR) and photo- reflective targets were attached to a tight-fitting headpiece or adhered to the skin overlying key skeletal landmarks. Each subject was randomly assigned to either the 60° or 90° direction and was exposed to 4 test conditions - arms up (with hands on their knees) with EMSR on, arms down (with hands low on the hips) with EMSR on, arms up with EMSR off, arms down with EMSR off. The effect of age and pretensioning on the following outcomes was quantified: 1) lateral and forward displacement of the torso, 2) torso rollout angle projected onto three orthogonal planes, and 3) resultant belt-sternal distance. The effect of pretensioning on torso containment within the shoulder belt was strong in both impact directions across all metrics evaluated. EMSR activation significantly reduced lateral displacement of the suprasternal notch (~100 mm, p<0.0001), coronal projection of the torso rollout angle (~15°, p<0.0001), and belt sternal distance when the arms were down (~50 mm, p<0.05). The benefit of pretensioning was achieved by early engagement of the torso by the shoulder belt. An added benefit of pretensioning was the ability to make similar the torso kinematics across a range of anthropometries as assessed within and across age groups. These results can serve as a data set for validating the responses of restrained ATDs and computational human models to low severity far side collisions, in particular the interaction between the torso and the shoulder belt.