BACKGROUND: Backward walking speed is an objective index to evaluate motor ability, but it is not clear whether it can screen stroke patients with mobility disorders. This study aims to test the diagnostic efficacy of backward maximum walking speed (BMWS) and backward comfortable walking speed (BCWS) on mobility defects in stroke patients and obtain the cutoff values of them. METHODS: We conducted a cross-sectional study of poststroke patients who were hospitalized from July 2021 to January 2022. 98 poststroke patients were selected for 6 functional tests and divided into the high-mobility function group and low-mobility function group by K-means cluster analysis. According to the test results of BMWS and BCWS, the receiver operating characteristic curve was drawn to determine the cutoff values and compare the diagnostic efficiency of the 2 indexes. RESULTS: The BMWS of stroke patients was 0.54 (0.29) m/s, and the BCWS was 0.37 (0.19) m/s. Receiver operating characteristic results showed that the cutoff value of BMWS for the diagnosis of mobility impairment in stroke patients was 0.3 m/s, the area under the curve was 0.95 (95% confidence interval, 0.89-0.98), the specificity was 0.86, and the sensitivity was 0.93. The cutoff value of BCWS was 0.27 m/s, the area under the curve was 0.91 (95% confidence interval, 0.84-0.96), the specificity was 0.91, and the sensitivity was 0.8. There was no significant difference between the 2 indexes in the diagnosis of mobility defects in stroke patients ( P > .05). CONCLUSIONS: Backward walking speed is a useful indicator for discriminating stroke patients with mobility deficits. Considering the risks associated with backward walking, BCWS can be used in clinical assessments to guide nurses in the development and implementation of rehabilitation programs.