The major advantages of radio frequency identification (RFID) technology over barcodes are that the RFID-tagged objects do not require to be in line-of-sight with the reader for their identification and multiple objects can be read simultaneously. But when multiple objects are read simultaneously there is always a problem of collision which reduces the efficiency of the system. This thesis presents a comprehensive study of the dynamic framed slotted ALOHA (DFSA)-based anti-collision multi-tag identification algorithms for passive RFID system. Performance of various DFSA algorithms is compared through extensive simulation results. In addition, a number of simple performance improvement techniques have also been investigated in this thesis, including improved estimation techniques for the number of tags in each read cycle and a low-complexity heuristic stopping criterion that can be easily implemented in the practical system.