Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder known by aberrant functional and structural brain connectivities. The main outcome of this “dysconnectivity” is usually a triad of repetitive behaviors, underdeveloped social skills, and delayed or no language development.[1] Underlying causes of ASD are still poorly understood and there have been efforts to identify the brain pathology in this disorder[2]. One of the promising approaches is the use of connectivity matrices generated by functional or structural brain MRIs to explore graph metrics. Despite their usability, these approaches are subject to several problems including family wise error (FWE) in multiple comparisons[3]. One of the provided solutions is Network-based statistics (NBS)[4], which is a permutation based thresholding of graph matrices and applies a weak control over FWE. Moreover, functional and structural connectivity relationship is one of the recent interests in brain imaging studies[5]. In this study, we used NBS to identify functionally connected nodes, then we compared structural graph metrics in these nodes to see function/structure relationship in ASD.We used an online open-source connectivity database (umcd.humanconnectomeproject.org)[6]. “UCLA_Autism” study connectivity matrices were used[7]. This study contains both structural and functional connectivity matrices for 69 subjects (35 ASD, 34 Typically Developing “TD”). The atlas used was generated by Power et, al. This atlas is very sensitive and specific to functional MRI connectivity signals and can be a better predictor of true functional connectivity[8]. These data are matched for age, sex and IQ. Structural matrices were number of fibers between two regions. The weighted structural matrices were built, as a number of fibers connecting two labels divided to all fibers in that network[9]. We used z-transformed functional matrices in NBS with 10000 permutations and t-threshold of 3.4 for two contrasts: ASD>TD and TD>ASD. The resultant significant nodes of functional matrices were then fed into Brain Connectivity Toolbox (BCT)[10] to calculate Local Efficiency (which is a measure of network segregation) from structural matrices. The results were analyzed with R v3.2.2 (https://www.r-project.org/) and corrected for false discovery rate (FDR).NBS showed a network of significant increased functional connectivity in ASD compared to TD (fig. 1, p-value=0.033). This network is comprised of 23 nodes and edges mostly connecting frontal and posteroinferior temporal cortices. On the other hand, in TD, NBS revealed a subnetwork of increased functional connectivity comprised of edges mostly connecting default-mode network (DMN) of the brain (e.g. precuneus, cingulate gyrus, angular gyrus; fig. 2, p-value=0.030). Our post-hoc analysis revealed no significant differences in local efficiency in the selected nodes.ASD is a neurodevelopmental disorder characterized by behavioral and cognitive problems[1]. Despite medical advances, the pathophysiology of this disease is still poorly understood. In this study we used NBS to investigate functional connectivity in ASD comparing to TD. Our study showed functional dysconnectivity in ASD. We found that frontal and posteroinferior temporal cortices’ show abnormally left-sided increased functional connectivity. This network is largely associated with social cognition and increased connectivity in this network may somehow explain the social problems in ASD, although studies relating social impairment and this network are required[11]. Moreover, patients with ASD showed reduced connectivity in DMN. DMN is anti-correlated with task-related networks and is believed to be involved in task-independent self-introspection[12]. Our finding is consistent with many studies indicating decreased functional connectivity in DMN, which has been related to symptoms of ASD[13]. We offer a new approach to study brain functional and structural connectivity. One of the solutions for family-wise errors is using the candidate nodes, i.e. choose limited number of nodes based on a hypothesis[3]. We propose a systematic approach, using NBS to identify significantly differing functionally nodes and investigate structural graph metrics in those to have brain function and structure in one view, although we didn’t find significant results. In conclusion, NBS reveals abnormally increased and decreased connectivity in frontotemporal network and DMN, respectively, which may be contribute to ASD’s clinical presentations.