Abstract: Patients with depression usually show depressive state, accompanied with cognitive and physical symptoms. Functional near-infrared spectroscopy (fNIRS) was utilized to investigate 8-minute spontaneous hemodynamic activity in the prefrontal cortex of 28 patients with depressive episode and 30 healthy controls. By adopting a graph-theoretical and topological analysis, the results indicated that patients with depressive state showed altered network patterns in the prefrontal cortex with significantly reduced density (lower average node degree), less cliquishness (lower average clustering coefficient), poor efficiency (higher average path length) and greater randomness (weaker small-world properties). These results support fNIRS measurement of the prefrontal cortex networks in the resting state as a feasible and effective way to assess the neural properties of psychiatric disorders.