Backward walking training has been reported to improve gait speed and balance post-stroke. However, it is not known if gains are achieved through recovery of the paretic limb or compensations from the nonparetic limb. The purpose of this study was to compare the influence of backward locomotor training (BLT) versus forward locomotor training (FLT) on gait speed and dynamic balance control, and to quantify the underlying mechanisms used to achieve any gains. Eighteen participants post chronic stroke were randomly assigned to receive 18 sessions of either FLT (n = 8) or BLT (n = 10). Pre- and post-intervention outcomes included gait speed (10-meter Walk Test) and forward propulsion (time integral of anterior-posterior ground-reaction-forces during late stance for each limb). Dynamic balance control was assessed using clinical (Functional Gait Assessment) and biomechanical (peak-to-peak range of whole-body angular-momentum in the frontal plane) measures. Balance confidence was assessed using the Activities-Specific Balance Confidence scale. While gait speed and balance confidence improved significantly within the BLT group, these improvements were associated with an increased nonparetic limb propulsion generation, suggesting use of compensatory mechanisms. Although there were no improvements in gait speed within the FLT group, paretic limb propulsion generation significantly improved post-FLT, suggesting recovery of the paretic limb. Neither training group improved in dynamic balance control, implying the need of balance specific training along with locomotor training to improve balance control post-stroke. Despite the within-group differences, there were no significant differences between the FLT and BLT groups in the achieved gains in any of the outcomes.