Radio-frequency identification (RFID) is a noncontact automatic identification technology that has been widely used in various modern industries. Due to the continuous improvement in sensing capabilities, energy autonomy, affordability, and usability, RFID is one of the key enabling technologies in realizing the Internet of Things (IoT). With their non-line-of-sight wireless transmission, wireless power, and sensing capability, lightweight RFID sensor tags can be connected to people or objects, which is crucial for future IoT applications such as healthcare, logistics, and manufacturing. The RFID approach with the simultaneous reading of multiple tag features can quickly identify connected “things” for further data communication and integration. Reducing the identification processing time for many tags located inside the interrogation range of the reader is essential. Therefore, an effective protocol must be developed to prevent collisions and reduce useless queries between the reader and tags to achieve rapid identification. Predetection tree-based algorithms, which are based on certain small predetection timeslots that tags can respond to, can eliminate unnecessary idle slots and prevent collision slots. In this paper, an efficient predetection-based anticollision algorithm has been proposed to achieve outstanding tag identification performance. The numbers of collision slots and idle slots are reduced by exploiting a novel predetection mechanism and slot size adjustment technique. Simulation results indicate that the proposed scheme reduces the tag identification time by approximately 30–60% compared with previously proposed predetection-based protocols. Moreover, the numbers of query cycles, collisions, and slots are smaller in the proposed scheme than in the previously proposed protocols.