A new design of the collision prevention system based on cascaded fuzzy inference system (CFIS) for lane-changing maneuver and car-following is proposed. The lane-changing is combined with car-following to save the processing time of the lane-changing maneuver. The distance and speed of the host vehicle (HV) relative to the vehicles in front or on target adjacent lane are measured by the spread spectrum radars and applied to the collision prevention system. The collision prevention system provided the output acceleration or deceleration rate for the host vehicle is designed according to the characteristics of the vehicles. Monte Carlo simulations were conducted to demonstrate that collision prevention system for car-following and lane-changing can provide a safe, reasonable and comfortable driving. The required processing time to complete the lane changing and to achieve the safe distance between the HV and the leading vehicle (LV) on target adjacent lane is about 16 seconds. The mean required time for lane-changing maneuver is three seconds, which is very close to the road test results.