Objective To find a simple, sensitive and effective method to predict the occurrence of cognitive impairment in patients after a transient ischemic attack (TIA). Methods Thirty-six persons who had survived a first TIA and could walk independently were selected into a TIA group, with another 36 healthy counterparts chosen to form a control group. Those in both groups were given a gait analysis during dual-task walking at the outset and then assessed using the MoCA scale 12 months later. Statistical analyses were conducted to assess the reliability of gait parameters in predicting cognitive dysfunction, and the area under the receiver operating characteristics curve was calculated. Results On the day after enrollment the variation in step length among the TIA patients with an abnormal step length was compared with variation among the normal controls and among the TIA patients with non-abnormal step-size. The differences were significant. The average walking speed of the TIA patients with an abnormal average walking speed was significantly different from that of the control group and that of the TIA patients with non-abnormal average speed. On the day after enrolling, the accuracy rate of the control group in the dual-task walking test was significantly better than that of any of the TIA patients, and the accuracy rate of the TIA patients with an abnormal gait in the dual-task walking test was significantly lower than that of the patients with a non-abnormal gait. Twelve months later the number of patients whose MoCA scores were less than 26 in the abnormal gait group was compared with that in the normal gait group, and the difference was statistically significant. The area under the receiver operating characteristics curve suggests that both the incidence of step length variation and the average stepping speed could significantly predict the occurrence of cognitive dysfunction 12 months later. Conclusion Quantitative analysis of gait while dual-task walking can effectively predict the occurrence of cognitive impairment in patients with TIA, and can detect abnormalities earlier than the MoCA scale test. Key words: Dual-task walking; Gait analysis; Montreal Cognitive Assessment; Cognition; Ischemia