Recent works have shown that, even in resting state, functional networks undergo dynamic changes over short time. In this study, we describe an approach to assess the difference in default mode network (DMN) dynamics between healthy controls (HC) and schizophrenia patients (SZ) using resting-state functional magnetic resonance imaging. Firstly, dynamic DMN was computed using a sliding time window method. Then, stability of the dynamic DMN evaluated using the spectrum of time-varying functional connectivity was compared between HC and SZ. Subsequently, the overall functional connectivity pattern and dynamic graph measures were also investigated for both groups. Results show that dynamic DMN of HC had more stable and stronger functional connectivity than that of SZ. Regarding to dynamic graph measures, SZ had lower connectivity strength, clustering coefficient, global efficiency, and local efficiency than HC. The findings suggest that dynamic functional network analysis is a promising technique for understanding schizophrenia.