ObjectivesThe aim of this study was to investigate the functional networks in subjects with reversible cerebral vasoconstriction syndrome (RCVS) using resting‐state functional magnetic resonance imaging (rs‐fMRI).MethodsWe prospectively recruited patients with RCVS and healthy controls (HCs) between February 2017 and April 2021. The rs‐fMRI data were analyzed using graph theory methods. We compared node‐based global and regional topological metrics (Bundle 1) and network‐based intranetwork and internetwork connectivity (Bundle 2) between RCVS patients and HCs. We also explored the associations of clinical and vascular (ie, the Lindegaard index, LI) parameters with significant rs‐fMRI metrics.ResultsA total of 104 RCVS patients and 93 HCs were included in the final analysis. We identified significantly decreased local efficiency of the left dorsal anterior insula (dAI; p = 0.0005) in RCVS patients within 30 days after disease onset as compared to HCs, which improved 1 month later. RCVS patients also had increased global efficiency (p = 0.009) and decreased average degree centrality (p = 0.045), clustering coefficient (p = 0.033), and assortativity values (p = 0.003) in node‐based analysis. In addition, patients with RCVS had increased internetwork connectivity of the default mode network (DMN) with the salience (p = 0.027) and dorsal attention (p = 0.016) networks. Significant correlations between LI and regional local efficiency in left dAI (r_s = −0.418, p = 0.042) was demonstrated.InterpretationThe significantly lower local efficiency of the left dAI, suggestive of impaired central autonomic modulation, was negatively correlated with vasoconstriction severity, which is highly plausible for the pathogenesis of RCVS. ANN NEUROL 2023;94:772–784