Abstract: Event-related potential (ERP) measurements are used to build functional network of spatial attention. The clustering coefficient is picked for analyzing this complex network. One new statistical parameter of existing edges percent between paired regions of interest (ROI) is proposed for analyzing ERP networks. Upon this,the properties of ERP functional network and the influences of locations of attention and stimulus are investigated.The fact that the clustering coefficient of ERP network is bigger than that of equivalent random network demonstrates the small world property of ERP network. Comparing existing edges percent between four ROI, the result shows that more edges exist between the stimulus contralateral posterior and anterior brain regions than those in ipsilateral regions. The statistical parameters of ERP networks between attention and unattention are obviously different, which indicates these parameters might be important indices of reflecting the ongoing brain dynamics. Proposal of new statistical parameters of complex networks may be a useful approach to study detailedly the connectivity of brain in various cognitive tasks.