The propose of the study was to evaluate the efficacy of the partial body weight support during treadmill training on the ambulation in elderly with chronic stroke. Fourteen hemiplegic volunteers participated and were divided into an experimental and control groups. In the experimental group, the body weight support during treadmill training was performed 3 times per week for 6 weeks. In the control group, usual treadmill training was applied. Before and after experiments, temporal-spatial gait parameters were measured. The date of 14 patients who carried out the whole experimental course were statistically analyzed. The results of the study were : 1. In the comparison of gait velocity before and after experiment, the gait velocity was significantly increased in the experimental group and the control group(p<.05). In the comparison of difference of the gait velocity between groups, there was not, significant difference between the experimental group and the control group(p>.05). 2. In comparison of gait cadence before and after experiment, the gait cadence was significantly increased in both groups(p<.05). In the comparison of difference of the gait cadence between groups, there was not significant difference between the experimental group and the control group(p>.05). 3. In the comparison of step length before and after experiment, the step length was significantly increased in the experimental group and the control group(p<.05). In the comparison of difference of the step length between groups, there was not significant difference between the experimental group and the control group(p>.05). 4. In the comparison of single support time asymmetry before and after experiment, the single support time asymmetry was no significant difference between groups(p>.05). In the comparison of difference of the single support time asymmetry between groups, there was not significant difference between the experimental group and the control group(p>.05). 5. In the comparison of step length asymmetry before and. after experiment, the step length asymmetry was not significant difference between the experimental group and the control group(p>.05). In the comparison of difference of the single step length asymmetry between groups, there was not significant difference between the experimental group and the control group(p>.05).