This is a single blind randomized controlled trial to examine the effect of virtual reality-based training on the community ambulation in individuals with stroke. Twenty subjects with stroke were assigned randomly to either the control group (n = 9) or the experimental group (n = 11). Subjects in the control group received the treadmill training. Subjects in the experimental group underwent the virtual reality-based treadmill training. Walking speed, community walking time, walking ability questionnaire (WAQ), and activities-specific balance confidence (ABC) scale were evaluated. Subjects in the experimental group improved significantly in walking speed, community walking time, and WAQ score at posttraining and 1-month follow-up periods. Their ABC score also significantly increased at posttraining but did not maintain at follow-up period. Regarding the between-group comparisons, the experimental group improved significantly more than control group in walking speed (P = 0.03) and community walking time (P = 0.04) at posttraining period and in WAQ score (P = 0.03) at follow-up period. Our results support the perceived benefits of gait training programs that incorporate virtual reality to augment the community ambulation of individuals with stroke.