Abstract: Abstract: In this paper, the water flow mechanism is introduced to improve the performance of artificial fish swarm algorithm (AFSA). Water flow mechanism simulates the natural phenomenon of running water flow in rivers, including persistent water flow and periodic water flow. Persistent water flow affects fish’s physical power continuously, which is utilized to adjust the visual and step parameters self adaptively. Periodic water flow rushes fish swarm effectively and changes the position of some fishes. Therefore, water flow mechanism can improve the search precision and maintain the diversity of population. The experiment results via benchmark function illustrate the effectiveness of the improved algorithm.