Radio frequency identification (RFID) provides a contactless approach for object identification. If there are multiple tags in the interrogation zone of a reader, tag collision occurs due to radio signal interference. To solve tags identification collision and improve identification efficiency in RFID system, a flood division anticollision (FDAC) algorithm has been presented. Firstly, the algorithm launches an estimation of the number of tags and according to the estimation result decides whether a flood diversion processing needs to be started or not. Secondly, when the flood diversion processing needs to be done, all tags are grouped and assigned to different models in which the tags are to be processed and identified in parallel. Thirdly, in the identification processing, for reducing data transmission, the reader needs only to send a three-dimensional-vector command to tags, tags respond to the command with part of collision-bit parameters, and stack and queue are adopted to store precious request command and tags' ID to avoid repeatedly transmitting them between the reader and the tags. Simulation experiment results show that FDAC is superior to the dynamic frame slotted (DFS) Aloha algorithm, the binary-retreat tree algorithm (BRT) and the dynamic binary-search tree (DBST) algorithm, in the performances of data bits transmission, identification time delay, and energy consumption by the reader.