Abstract: To optimize the performance of radio frequency identification system (RFID), achieve efficient management and ensure fishery production safety, the paper put forward an improved dynamic frame slotted ALOHA anti-collision algorithm based on the analysis of the existing ALOHA algorithm. The algorithm changed the frame slot length according to the number dynamics of tags. The improved algorithm used the dynamic frame slotted algorithm to estimate the number of tags. Simulation results showed that the maximum throughput of the improved algorithm system reached 43% when the number of tag increased or decreased. The throughput changed little when the number of tags changed. Compared with the traditional algorithms, the improved algorithms had obvious advantages on response time of the system when the number of tags changed much, and the response time reduced almost in half. The throughput and response delay of the improved algorithm had been significantly enhanced compared with the traditional algorithm. The new algorithm improved the performance of radio frequency identification system and better adapted to the requirements of the system.