Background <p>Internet gaming disorder (IGD) is associated with abnormal functional connectivity (FC) in brain networks. However, findings from resting-state functional magnetic resonance imaging studies are highly inconsistent, likely due to individual heterogeneity in IGD-related neural alterations—a feature commonly observed in other psychiatric disorders but understudied in IGD.</p> Methods <p>We applied normative modeling to nucleus accumbens (NAcc) seed-to-voxel resting-state FC to derive individualized deviation (Z) maps for 173 IGD participants relative to age- and sex-adjusted normative ranges from 232 healthy controls. We then performed exploratory unsupervised clustering of network-level deviation features across three sample data, three atlas templates, and two clustering algorithms, selecting the optimal number of clusters using the silhouette criterion.</p> Results <p>IGD showed marked heterogeneity in FC deviations: voxel-level deviations were largely idiosyncratic in both spatial distribution and direction. When deviations were summarized at the network level, clustering consistently selected a two-cluster solution across data, atlases, and algorithms, separating a majority “low-deviation” stratum from a smaller “high-deviation” stratum.</p> Conclusions <p>IGD is characterized by pronounced individual variability in FC deviations. Network-level deviations yield a robust higher- vs. lower-deviation stratification, although the present findings do not support interpreting this pattern as evidence for discrete subtypes. The present study highlights the utility of individualized deviation mapping beyond conventional case–control analyses for characterizing heterogeneity in IGD.</p>