Audiovisual working memory (WM) plays a critical role in multisensory cognitive processing, yet its structural neural correlates remain insufficiently understood. This study employed an audiovisual dual n-back task paradigm and voxel-based morphometry (VBM) to investigate gray matter volume (GMV) associations with behavioral performance in 60 healthy individuals. Behavioral results revealed a significant visual dominance effect under high cognitive load: visual performance remained stable across conditions, whereas auditory performance declined. Structural analyses showed modality-specific GMV correlations. Visual performance was positively associated with GMV in the insula, posterior cingulate, hippocampus, and inferior frontal regions, while auditory performance was negatively correlated with GMV in the angular and middle occipital gyri. Notably, the left cuneus exhibited a strong positive correlation with the Δd prime difference under high load, suggesting its potential role in cross-modal resource allocation. Furthermore, cognitive overload appeared to disrupt the structure-behavior associations observed under lower load, highlighting a load-dependent dissociation within executive control and sensory integration regions. These findings underscore the distinct anatomical substrates supporting audiovisual WM and the neural basis of visual dominance, offering structural markers for targeted cognitive training and clinical intervention.