The current anti-collision algorithms for Radio Frequency IDentification (RFID)was analyzed and compared, and an optimized mixed frame anti-collision algorithm based on the binary tree and frame time slotted ALOHA was presented to resolve the problems in traditional algorithms. In proposed algorithm, the frame time slot algorithm was used to identify collision time slot when a reader communicated with the labels. However, if the collision labels have happened in the one time slot, the binary tree algorithm was used to make a clear distinction. By using the Manchester decoder to get collision location, the states of collision was estimated in one time slot, then the mathematical expectation was obtained by using mathematical deduction and R software programming. The simulation shows that the proposed mixed algorithm gives full play to the advantages of the two algorithms, its throughput rate, searching time and transmission delay are all superior to that of the traditional algorithm, especially, the throughput rate is nearly higher 50% than that of the ALOHA algorithm. Moreover, the data transmission shows higher security by decreasing of the burden of other transport.