Objective The aim of the study is to determine whether virtual reality-based dynamic standing balance training improves three elements of sensory integration and investigate whether virtual reality-based dynamic standing balance training results in improved outcomes, especially regarding balance and gait, compared with the standard training method. Design This single-blinded, randomized, controlled trial involved 30 patients with hemiplegia. The experimental (n = 15) and control (n = 15) groups received virtual reality augmented-standing balance training or standard standing balance training, respectively, for 20 mins, 5 days a week, for 3 wks. The patients were assessed for primary (Sensory Organization Test and the Berg Balance Scale) and secondary (the functional reaching test and timed up-and-go test) outcomes before and after training. Results From preintervention to postintervention, the Berg Balance Scale score (F = 26.295, P < 0.05), timed up-and-go test score (F = 18.12, P < 0.05), mean score of conditions 2 (F = 4.36, P < 0.05) and 6 (F = 5.61, P < 0.05), and composite score of the Sensory Organization Test (F = 5.385, P < 0.05) in both groups were significantly improved. However, there was no significant difference between experimental group and control group (time*group P > 0.05). Conclusions Virtual reality combined with standing balance training improved sensory integration, postural control, balance, and gait ability in patients with hemiplegia, reducing fall risk. However, outcomes were comparable with general balance training regarding balance and gait.