In response to issues such as the inability of EEG rhythm energy to reflect temporal information and insufficient exploration of spatial information, this paper uses microstate analysis methods to study the spatiotemporal patterns of electroencephalogram (EEG) related to virtual reality motion sickness (VRMS), in order to detect VRMS. Using Multivariate Variational Mode Decomposition (MVMD), EEG signals were divided into 5 frequency bands from low frequency to high frequency. The frequency, average duration, coverage, and conversion rate of EEG microstates were analyzed, and the effectiveness of these features was verified using statistical analysis and classification methods. The research results show that the classification accuracy of fusing all features in 5 frequency bands reaches the maximum value of 83.9%. Therefore, microstate methods are expected to provide new ideas for studying VRMS.