<p>Balance training in geriatric rehabilitation reduces fall risk by eliciting multifaceted cortical reorganization through repeated sensorimotor challenges, supporting more efficient postural control in aging. This study combined minimum spanning tree (MST) and k-iteration second-best MST analyses to characterize training-induced changes in cortical network efficiency and resilience. Twenty-four older adults (70.4 ± 3.3&#xa0;years) completed 12 sessions of stabilometer training with real-time visual feedback. Pre- and post-test assessments evaluated EEG-derived backbone connectivity and postural fluctuation dynamics during goal-directed stance. Post-training, participants exhibited reduced sway magnitude (<i>p</i> = 0.004), increased signal complexity (<i>p</i> = 0.009), and higher movement frequency (<i>p</i> = 0.045), indicating enhanced postural regulation. MST analysis revealed a shift toward a more integrated, hierarchically organized network, evidenced by decreases in kappa (<i>p</i> = 0.011) and leaf fraction (<i>p</i> = 0.014), alongside an increase in maximum betweenness centrality (<i>p</i> = 0.005). Iterative second-best MST analysis further showed that relative total connectivity changes during the 15th–50th iterations were less negative post-training (<i>p</i> &lt; 0.05), whereas algebraic connectivity (λ₂) in the 1st–5th iterations declined significantly (<i>p</i> &lt; 0.05), indicating enhanced peripheral flexibility and consolidated core structure. In conclusion, short-term postural training induced layered cortical network reorganization that is mechanistically consistent with more efficient postural regulation in older adults.</p>