<p>Post-stroke depression (PSD) is a common complication among stroke survivors and is closely associated with executive function deficits, yet effective treatments targeting both domains remain limited. Transcutaneous auricular vagus nerve stimulation (taVNS) has demonstrated antidepressant effects in preliminary trials; however, the neural mechanisms through which it modulates mood and cognition in PSD are poorly understood. This study aimed to investigate the efficacy of taVNS on depressive symptoms and executive function in patients with PSD and to elucidate the underlying neural mechanisms using functional near-infrared spectroscopy (fNIRS). In this sham-controlled, assessor-blinded randomized trial, 32 patients with PSD were randomly assigned to receive either active taVNS (left cymba conchae, 30&#xa0;Hz, 30&#xa0;min twice daily; <i>n</i> = 16) or sham stimulation (<i>n</i> = 16) for 4 weeks, alongside conventional rehabilitation. The primary outcome was the Hamilton Depression Rating Scale (HAMD-17). Secondary outcomes included the Modified Barthel Index (MBI) and Stroop Color–Word Test (SCWT). Task-related cortical activation was analyzed using linear mixed models, and resting-state functional connectivity was examined using network-based statistics (NBS). A significant group × time interaction was observed for HAMD-17 scores (F = 11.680, <i>P</i> = 0.002), with the taVNS group showing a significantly greater reduction in depressive symptoms. Task-state fNIRS revealed significant interaction effects in the left dorsolateral prefrontal cortex (DLPFC) and left pre-supplementary motor area (pSMA) (FDR-corrected <i>P</i> &lt; 0.05). Resting-state NBS revealed enhanced frontal–motor connectivity within the taVNS group, involving bilateral DLPFC, OFC, and pSMA, while between-group comparison showed a right-hemisphere-dominant lateralization pattern. No significant group × time interaction was found for the MBI or for any SCWT interference index (reaction time, accuracy, or inverse efficiency score). In exploratory analyses within the taVNS group, increases in left DLPFC activation were associated with greater reductions in HAMD-17 scores (<i>r</i> = − 0.646, <i>P</i> = 0.007), and increases in left pSMA activation were associated with greater reductions in SCWT reaction-time interference (<i>r</i> = − 0.736, <i>P</i> = 0.001). taVNS reduced depressive symptoms more than sham stimulation in patients with PSD. Although behavioral evidence for executive function benefit remained inconclusive, fNIRS findings suggested treatment-related changes in cortical activity and connectivity within executive-control-related regions, particularly the prefrontal–pre-supplementary motor area network. These findings provide preliminary neuroimaging evidence for taVNS as a neuromodulatory intervention for PSD, though replication in larger, multicenter trials is warranted.</p><p>Trial registration: ChiCTR2400084576 (Registration date 21/05/2024).</p>