The increasingly busy maritime traffic system has resulted in frequent maritime accidents. To prevent maritime accidents, we investigated and studied typical cases of maritime accidents along the coast of China in recent years. Based on the complex network theory, a network model of China's coastal maritime traffic accidents is constructed from the four risk factors of human-ship-environmental-management. On this basis, the overall structural characteristics of the maritime accident network are analyzed according to the characteristic parameters of the complex network. In addition, based on the node betweenness and PageRank value, the network robustness analysis is carried out, the key factors in the maritime accidents are excavated and risk prevention and control measures are proposed. The research shows that the average path length of the marine traffic accident network is short, the clustering coefficient is high, and the risk spreads quickly in the network and passes through fewer nodes, which can lead to the occurrence of accidents. Through the management and control of key risk factors, the evolution of risks can be effectively blocked and the large and above maritime accidents can be reduced.