Background: Examining between-sessions test-retest reliability of functional near-infrared spectroscopy (fNIRS) data is crucial to better interpret rehabilitation-related changes in the hemodynamic response. Research question: This study investigated test-retest reliability of prefrontal activity during usual walking in 14 patients with Parkinson's Disease with a fixed retest intervals of five weeks. Methods: Fourteen patients performed usual walking in two sessions (T0 and T1). Relative changes in cortical activity (oxy and deoxyhemoglobin: ΔHbO 2 and ΔHbR, respectively) in the dorsolateral prefrontal cortex (DLPFC) using fNIRS system and gait performance were measured. Test-retest reliability of mean ΔHbO 2 for the total DLPFC and for each hemisphere were measured using paired t-test, intraclass correlation coefficient (ICC), and Bland-Altman plots with 95% agreement. Pearson correlations between cortical activity and gait performance were also performed. Results: Moderate reliability was found for ΔHbO 2 in the total DLPFC (mean difference of ΔHbO 2 between T1 and T0 = − 0.005 µmol, p = 0.93; ICC average = 0.72). However, test-retest reliability of ΔHbO 2 was poorer when considering each hemisphere. Significance: Findings suggest that fNIRS may be used as a reliable tool for rehabilitation studies in patients with PD. Test-retest reliability of fNIRS data between 2 sessions during walking tasks should be interpreted respectively of gait performance.