Forward safety system (FSS) monitors the roadway ahead and warns the driver when a potential risk exists. The vision-based FSS can provide better protection capability than those based on radars because it owns rich environmental information, such as the shape and color of lane lines and objects. However, the detection of vision-based FSS is generally restricted to a small area in front of the vehicle due to the hardware limitation of the adopted camera. This paper presents a novel FSS which utilizes cameras with different focal lengths to extend the detection range and thus can provide early lane departure and forward collision warning. The proposed FSS extracts lane lines by the intensity and structural properties, and detects potential lane departure and forward collision by analyzing the moving trajectory and by detecting the obstacle ahead, respectively. Our experimental results showed the FSS, which uses two cameras with detection range from 5 m to 120 m in this work, achieved high accuracy in both the lane departure warning junction (96%) and the forward collision warning function (98%). The average accuracy of the object distance reported in the FSS is above 95%.