Background: In the stroke patients with the characteristics of hemiplegic gait, turning direction of the affected and unaffected side influences turning time. Therefore, it is important to investigate the walking response to turning directions in stroke patients. Objects: This study aimed to measure the walking time while turning direction in hemiplegic patients depending on balance ability measured by Berg Balance Scale. Methods: A group of forty-five subjects with stroke (Berg Balance Scale score≥46 were twenty-eight, Berg Balance Scale score≤45 were seventeen) were enrolled in this study. Subjects were asked to perform the Timed Up and Go test. Testing indications included two directions for turning in each subject. These indications were for turning toward the affected and unaffected side in stroke patients. The duration of total analysis duration, sit to stand phase, stand to sit phase, mid-turning phase, and end turning phase were recorded. The obtained data were analyzed by using paired t-test and Wilcoxon signed rank test in the group that are below and above 45 points of Berg Balance Scale score. The significance level was set at ɑ=.05. Results: There were significant increase time in the analysis duration and end turning phase duration while subjects were turned the unaffected side in stroke patients that presented a Berg Balance Scale score≤45 (p<.05). However, the comparison between the affected side and the unaffected side in the stroke patients with Berg Balance Scale score≥46, revealed no significant differences of the measured parameters. Conclusion: This finding should be suggested in the specific definition of turning direction for evaluation with Timed Up and Go test in the Berg Balance Scale score≤45, and other intervention for hemiplegic patients need to be suggested the direction of turning during walking training program.