As part of a larger gait adaptability training study, we designed a program that presented combinations of visual flow and support-surface manipulations to investigate the response of healthy adults to walking on a treadmill in novel discordant sensorimotor conditions. A visual dependence score was determined for each subject, and this score was used to explore how visual dependency was linked to locomotor performance (1) during three training sessions and (2) in a new discordant environment presented at the conclusion of training. Performance measures included reaction time (RT), stride frequency (SF), and heart rate (HR), which respectively served as indicators of cognitive load, postural stability, and anxiety. We hypothesized that training would affect performance measures differently for highly visually dependent individuals than for their less visually dependent counterparts. A seemingly unrelated estimation analysis of RT, SF, and HR revealed a significant omnibus interaction of visual dependency by session (p < 0.001), suggesting that the magnitude of differences in these measures across training day 1 (TD1), training day 3 (TD3), and exposure to a novel test is dependent on subjects' levels of visual dependency. The RT result, in particular, suggested that highly visually dependent subjects successfully trained to one set of sensory discordant conditions but were unable to apply their adapted skills when introduced to a new sensory discordant environment. This finding augments rationale for developing customized gait training programs that are tailored to an individual. It highlights one factor--personal level of visual dependency--to consider when designing training conditions for a subject or patient. Finally, the link between visual dependency and locomotor performance may offer predictive insight regarding which subjects in a normal population will require more training when preparing for specific novel locomotor conditions.