PurposeThis study aimed to explore the test-retest reliability of fNIRS in measuring frontal and parietal cortices activation during straight walking and turning walking in older adults, in order to provide a theoretical foundation for selecting assessment tools for clinical research on motor control and some diseases such as Parkinson's disease in older adults. Methods18 healthy older participants (69.1 ± 0.7 years) were included in this study. The participants completed straight walking and figure-of-eight turning walking tasks at self-selected speeds. Intra-class correlation coefficients (ICCs) and Bland-Altman scatter plots were used to assess the test-retest reliability of oxyhemoglobin (HbO2) changes derived from fNIRS. p < 0.05 was considered statistically significant. ResultsThe test-retest reliability of HbO2 in prefrontal cortex (ICC, 0.67–0.78) was good and excellent, in frontal motor cortex (ICC, 0.51–0.61) and parietal sensory cortex (ICC, 0.53–0.62) is fair and good when the older adults performed straight and turning walking tasks. Bland-Altman diagram shows that the data consistency is fair and good. ConclusionfNIRS can be used as a clinical measurement method to evaluate the brain activation of the older adults when walking in a straight line and turning, and the results are acceptable repeatability and consistency. However, it is necessary to strictly control the testing process and consider the possible changes in the repeated measurements.