Background: Controversial findings exist in the literature with respect to the efficacy of visually guided weight-shifting (WS) training as a means of improving balance in healthy older adults. Objective: The purpose of this study was to investigate the impact of two direction-specific, visually guided WS training protocols on standing balance of healthy elderly women. Methods: Forty-eight community-dwelling elderly women, all free of any neurological or musculoskeletal impairment, were randomly assigned into: a group that practiced WS in the anterior/posterior direction (A/P group, n = 19), a group that practiced WS in the medio/lateral direction (M/L group, n = 15) and a control group (n = 14). Participants performed 12 training sessions of visually guided WS (3 sessions a week for 25 minutes per session). Static balance was measured before and after training in normal (bipedal) quiet stance (NQS) and sharpened-Romberg stance (SRS) by recording center of pressure (CoP) variations and angular segment kinematics. Results: In NQS, neither of the two training protocols had a significant impact on postural sway measures, although a significant decrease in interlimb asymmetry of CoP displacement was noted for the A/P group. In SRS, A/P training induced a significant reduction of CoP displacement, lower limb pitch and upper trunk roll rotation. Conclusion: The results of the study stress the importance of using direction-specific WS tasks in balance training, particularly in the A/P direction, in order to improve control of static balance in elderly women.