Annual Statistics on car accident" reveals that driver's error is one of the major causes of car accidents on curved roads and some of they can be classified into crashes on road shoulders. Car vendors have developed some lane keeping assist systems (LKAS) to prevent vehicles from straying into other lanes or straying off the road. These systems mainly use lane recognition schemes based on image processing and they suffer from poor visibility problems. To solve this problem, we are investigating a lane keeping assist system that uses RFIDs. We constructed a mathematical model that takes a vehicle's movement into consideration, and evaluated the effectiveness of the system based on this model. The evaluation showed that the system can give drivers effective warnings to prevent their cars from straying onto road shoulders if RFIDs are placed at 5-meter intervals on the shoulders.