Background: Gait problems appear in most stroke patients. Commonly, stroke patients show the typical abnormal gait patterns, such as circumduction, genu recurvatum, and spastic paretic stiff-legged gait. An inclined treadmill gait exercise is good for gait problems of stroke patients. In addition, the backward walking training has been recommended in order to improve the component of the movement for the forward walking. Objects: The purpose of this study to investigated the effects of backward walking with inclined treadmill training on the gait in chronic stroke patients. Methods: A total of 30 volunteers were randomly allocated to two groups that walked on an inclined treadmill: the experimental group ($n_1=15$), which walked backward, and the control group ($n_2=15$), which walked forward. To measure the improvement of the patients' gait, a Figure of Eight Walking Test (F8W), Four Square Step Test (FSST), and Functional Gait Assessment (FGA) were performed. We also measured spatio-temporal gait variables, including gait speed, cadence, stride length, and single limb support using a three-axial wireless accelerometer. The measurements were taken before and after the experiment. The Wilcoxon signed-rank test was used to compare both groups before and after the interventions. The Mann-Whitney U test was used for the comparisons after the interventions. The statistical significance was set at ${\alpha}=.05$. Results: Before and after experiment, all dependent variables were significantly different between the two groups (p<.05). As compared to the control group, the experimental group showed more significant improvements in F8W, FSST, speed, cadence, stride length, and single limb support (p<.05); however, FGA in this group was not significantly different from the control (p>.05). Conclusion: Our results suggest that backward walking on an inclined treadmill is more effective for improving the gait of stroke patients than forward walking.