SUMMARYRadio frequency identification (RFID), a new technology for the contactless recognition of objects, is considered an ideal solution to replace barcode systems. To make RFID technology widely applicable, it is important to resolve the ‘multiple tag identification problems’, that is, the problem of indicating tag collisions occurring when multiple RFID tags are simultaneously present in a reader's field. The proposed anti‐collision algorithm in this paper consists of two separate phases: distribution and recognition. The algorithm distributes tags to be read in such a way that tag collisions are reduced. In the distribution phase, the conventional ALOHA algorithm is used with a newly added command. When there are several occurrences of collision or no‐response in a round, a new round is initiated by adjusting the frame size on the basis of the estimated number of tags. ALOHA‐based algorithms have the weakness of not being able to guarantee a 100% tag identification rate. Hence, the proposed algorithm uses a tree‐based algorithm in the recognition phase in order to eliminate this problem. In addition, the reversed ID approach is employed to ensure efficient tag identification. The simulation results show that the proposed algorithm takes less tag identification time than the conventional algorithms, thereby increasing RFID system efficiency. Copyright © 2012 John Wiley & Sons, Ltd.