Abstract: As a key component of Marine ecosystems, the relative importance of fish in the food web directly affects the overall stability of the ecosystem. At present, several complex network centrality indicators such as Katz index and PPR index have been used to assess the importance of fish, though each focuses on different aspects. In order to ensure the applicability of different centrality indicators in characterizing the importance of fish and to comprehensively evaluate the impact of fish on food web stability, this study, systematically defines fish centrality indicators including fish degree centrality and fish information centrality, and propose a Multi-centrality Index Similarity Fusion (MISF) method. This method constructs a fish characteristic matrix containing multiple network centrality indicators, proposes fish similarity calculation based on cosine similarity, Euclidean distance and relative entropy fusion, and uses the entropy weight method to determine the weight of each index, ultimately generating a comprehensive ranking of fish importance. Based on the data of fish predation network in the Pearl River Estuary from 2016 to 2018, this paper carried out the assessment of fish importance. Experimental results show that compared with Katz index, PPR index and other methods, the proposed method can effectively evaluate the relative importance of fish in the food web. In addition, according to the different importance levels of fish in the food web, the differences in the impact of fish population fluctuation on the stability of the network were discussed. This study provides a new perspective for fish resource management.