Neurological disorder can cause functional network changes in white matter (WM). However, changes in the WM functional network in children with drug-resistant temporal lobe epilepsy (DRTLE) require further clarification. Therefore, we combine graph theory with resting-state functional magnetic resonance imaging (rs-fMRI) and T1-weighted imaging (T1WI) to investigate the topological features of the WM network in children with DRTLE, discover potential biomarkers, and understand the underlying neurological mechanisms. We included 91 children (43 with DRTLE and 48 healthy controls), acquiring structural and functional MRI data to construct WM functional networks. Graph theory was applied to evaluate topological differences and their correlation with onset age, disease duration and cognitive measures. A Support Vector Machine model classified individuals with DRTLE based on WM connectivity, with accuracy validated through leave-one-out cross-validation. The global topological properties of the WM network in children with DRTLE were altered, manifesting as an imbalance between global integration and segregation Local nodal efficiency changes in the association fibers exhibited reduced information transfer and centrality at several nodes. Conversely, commissural and projection fibers displayed increased network properties. Cognitive metrics correlated with nodal disturbances. The classification model achieved 73.6 % accuracy and an area under the curve (AUC) of 0.744. This indicates that the WM functional network in DRTLE presents with anomalies in the topological attributes, which are associated with cognitive impairments. The WM functional connectivity may serve as valuable indicators for clinical classification of the condition. The insights provided have augmented our understanding of the complex neurological mechanisms involved in epilepsy.Copyright © 2025. Published by Elsevier Inc.