BackgroundDual-task walking is a good paradigm to measure the walking ability of stroke patients in daily life. It allows for a better observation of brain activation under dual-task walking to assess the impact of the different tasks on the patient when combining with functional near-infrared spectroscopy (fNIRS). This review aims to summarize the cortical change of the prefrontal cortex (PFC) detected in single-task and dual-task walking in stroke patients.MethodsSix databases (Medline, Embase, PubMed, Web of Science, CINAHL, and Cochrane Library) were systematically searched for relevant studies, from inception to August 2022. Studies that measured the brain activation of single-task and dual-task walking in stroke patients were included. The main outcome of the study was PFC activity measured using fNIRS. In addition, a subgroup analysis was also performed for study characteristics based on HbO to analyze the different effects of disease duration and the type of dual task.ResultsTen articles were included in the final review, and nine articles were included in the quantitative meta-analysis. The primary analysis showed more significant PFC activation in stroke patients performing dual-task walking than single-task walking (SMD = 0.340, P = 0.02, I2 = 7.853%, 95% CI = 0.054–0.626). The secondary analysis showed a significant difference in PFC activation when performing dual-task walking and single-task walking in chronic patients (SMD = 0.369, P = 0.038, I2 = 13.692%, 95% CI = 0.020–0.717), but not in subacute patients (SMD = 0.203, P = 0.419, I2 = 0%, 95% CI = −0.289–0.696). In addition, performing walking combining serial subtraction (SMD = 0.516, P < 0.001, I2 = 0%, 95% CI = 0.239–0.794), obstacle crossing (SMD = 0.564, P = 0.002, I2 = 0%, 95% CI = 0.205–0.903), or a verbal task (SMD = 0.654, P = 0.009, I2 = 0%, 95% CI = 0.164–1.137) had more PFC activation than single-task walking, while performing the n-back task did not show significant differentiation (SMD = 0.203, P = 0.419, I2 = 0%, 95% CI = −0.289–0.696).ConclusionsDifferent dual-task paradigms produce different levels of dual-task interference in stroke patients with different disease durations, and it is important to choose the matching dual-task type in relation to the walking ability and cognitive ability of the patient, in order to better improve the assessment and training effects.Systematic review registrationhttps://www.crd.york.ac.uk/prospero/, identifier: CRD42022356699.