INTRODUCTION: Exergame-based training is emerging as the most effective exercise modality for improving cognition, yet its neural correlates remain largely unexplored. This study explored gray matter (GM) and white matter (WM) changes following the addition of ‘Brain-IT’ training to usual care in mild neurocognitive disorder (mNCD) and their associations with cognitive performance changes. METHODS: We included 41 participants with mNCD, randomized to either the intervention (‘Brain-IT’ training + usual care) or the control (usual care only) group. ‘Brain-IT’ is a holistic, individually tailored program for secondary mNCD prevention delivered through serious exergames. T1-weighted and diffusion tensor imaging data were analyzed via standard neuroimaging analysis pipelines (FreeSurfer, tract-based spatial statistics) to assess GM/WM volumes in predefined regions of interest and WM integrity at the voxel-to-voxel level. Intervention-related changes were explored via analyses of covariance, focusing on effect size estimates. One-sided bivariate correlation analyses explored associations between changes in brain structure and cognitive performance. RESULTS: Complete datasets from 30 study participants (72.0 ± 8.6 years; 27% females) were available. 87% of participants had biomarker-supported characterization of mNCD etiology– mostly Alzheimer’s (62%). Significant moderate to large effects (partial eta-squared = 0.109 to 0.187) on GM/WM volumes were observed in the right and total hippocampus, thalamus, and anterior cingulate cortex in favor of ‘Brain-IT’ training. Hippocampal and thalamic changes correlated with improvements in verbal delayed recall. Protective effects on WM integrity, which correlated with cognitive improvements, were also observed, mainly around the thalamic radiation and the corpus callosum. CONCLUSION: This is the first RCT showing that a co-designed, purpose-developed, and individually tailored exergame-based training may positively impact brain structures affected in mNCD, with potential associations suggestive of a causal link to cognitive improvements. Since hippocampal atrophy is a hallmark of Alzheimer’s disease with high prognostic value for disease progression, our observations may be a first indication of a potential disease-modifying role of ‘Brain-IT’ training. However, adequately powered and hypothesis-driven studies are needed to build on these initial exploratory findings and better understand the neurobiological effects of exergame-based training. TRIAL REGISTRATION: ClinicalTrials.gov (NCT05387057; date of registration: May 18, 2022): https://clinicaltrials.gov/ct2/show/NCT05387057. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s13195-025-01835-2.