Mentalizing, the ability of perspective-taking, underlies individuals’ healthy social functioning and is supported by a core brain network, including temporoparietal and prefrontal regions. Evidence suggests similar brain regions to be recruited during mentalizing already in early childhood (Fehlbaum et al.,2022), but the changes of functional correlates in correspondence to neural network formation (i.e., functional connectivity (FC)) have yet to be tested. Here we report data from 81 typically developing children (6-14 y, mean=10.06 y; 29 girls) who completed CAToon (Borbás et al.,2020), an fMRI mentalizing task. Group-wise whole-brain functional activation was tested (FWE-corrected at p<0.05) and FC between core mentalizing regions of interest was computed (p<0.05, FDR-corrected). Age effects were tested for whole-brain functional activation, group-level FC, and network strength. In line with past evidence, children activated bilateral temporoparietal junction and superior/middle temporal gyri, precuneus, and medial frontal areas during mentalizing (Fig.1A). Age was not related to the functional activation pattern observed. FC during mentalizing in children was characterized by strong local short-range connections within prefrontal and temporoparietal brain regions, respectively (Fig.1B). Children’s age predicted the emergence of long-range connections, evidenced by prefrontal-temporoparietal coupling (Fig.1C). Furthermore, mentalizing-related overall network connectivity strength significantly increased with children’s age (r(79) =0.33, p=0.003). We demonstrate that mentalization during childhood results in neural activation of core areas of perspective-taking previously described in adults. The here observed age-related FC network changes support developmental theories suggesting local-to-global connectivity emergence (Fair et al., 2009) paralleled by skill acquisition across childhood.