Objective We aimed to explore the symptom-herb relationships and the evolution law implied in the development of stroke treated by traditional Chinese medicine ( TCM). Methods Based on ancient books on TCM, a stroke symptom-herb network was constructed, and its temporal evolution was analyzed by network topology and node ranking feature calculation, community discovery, subnetwork similarity calculation, and other method. Results The changes in the structure and scale of the symptom-herb network of stroke and evolution analysis of the node sorting characteristics revealed the key symptoms and core herbs used to treat stroke in different historical periods, which confirmed the changes in the theoretical understanding of stroke in TCM and the gradual maturity of symptom differentiation and treatment. Subnetwork similarity analysis showed that stroke can be roughly divided into three categories: true stroke, analogous stroke, and acute stroke. Community discovery and evolutionary analysis uncovered symptom and drug combinations that were closely linked in different historical periods. Conclusion The temporal evolution analysis of complex networks serves as a new method for studying the complex relationship and evolution law between stroke-related symptoms and herbs used. The result can be used as a reference for combining the source and flow of stroke and clinical symptom differentiation and treatment. [ABSTRACT FROM AUTHOR]