Purpose: The purpose of this study was to identify whether task-related circuit exercise program combined with sensorimotor training for 4 weeks could improve the balance and gait in stroke patients. Method: Fifteen stroke patients who had agreed with the study were randomly divided into 3 groups categorized as task-related circuit exercise program combined with sensorimotor training group (experimental group 1, n=5), task-related circuit exercise program group (experimental group 2, n=5), and control subjects performed conventional physical therapy (control group, n=5). The balance and gait were assessed by BT-4 force platform system, Berg Balance Scale, 10meter Walk Test and Smart Step at before training and after training. Wilcoxon signed rank test was used to analyze change before and after intervention in intra-group. Kruskal Wallis H test, Mann-Whitney U test and Bonfferoni correction were used to analyze changes of all variables in inter-groups. Result: The experimental group 1 showed significant improvements in postural sway area, BBS scores, walking velocity and plantar pressures of affected foot, whereas the experimental group 2 showed significant improvements in BBS scores, and the control group were no significantly different in all variables following training. The changes of postural sway area and BBS scores in the experimental group 1 were significantly greater than them of the control group. The changes of postural sway area in the experimental group 1 was significantly greater than that of the experimental group 2. Conclusion: The result of this study suggest the task-related circuit exercise program combined with sensorimotor training is an effective intervention to improve balance and gait in stoke patients.