Purpose Previous studies have discovered different neuroimaging features in anti-NMDAR encephalitis associated with cognitive dysfunction. However, it is unknown whether there is a correlation between abnormal homotopic connectivity and cognitive impairment in anti-NMDAR encephalitis. We aim to explore the homotopic connectivity patterns of patients with anti-NMDAR encephalitis and their associations with clinical characteristics. Methods Resting-state functional magnetic resonance imaging (rs-fMRI) was performed on 29 patients with anti-NMDAR encephalitis and 26 healthy controls (HCs). Voxel-mirrored homotopic connectivity (VMHC) and multivariate pattern analysis (MVPA) were applied to analyze the imaging data. A correlation was also performed between aberrant brain regions and clinical parameters. Results Compared to HCs, the performance of alertness in the patient group was typically worse ( p  < 0.05). A significant decrease in VMHC was observed in many regions of the patients in comparison to HCs, including the cerebellar 6, para-hippocampal gyrus, insula, precuneus, and middle frontal gyrus ( p  < 0.001). The insula and middle frontal gyrus were found to show positive correlations with alertness. The MVPA method achieved a classification accuracy of 74.55% with a sensitivity of 82.76% and a specificity of 65.38% in discriminating patients from HCs. Conclusion Our findings indicate that interhemispheric functional imbalance may play a significant role in the pathophysiology of cognitive dysfunction in anti-NMDAR encephalitis. The MVPA results suggest that abnormal VMHC may play a crucial role in the identification of patients with anti-NMDAR encephalitis from HCs.