The path recognition for vision-guided intelligent vehicle is easily affected by environmental conditions such as illumination. Considering the width of the path in the captured image should be in a fixed range, the recognized path width could be used to determine whether the previously-used threshold was suitable or not. And then the threshold could be adjusted to a suitable value to recognize the path accurately. The process and method for acquiring the dynamic image threshold was presented, and the path recognition and tracking system was designed. The experiments showed the dynamic image threshold could recognize the path more efficiently than other common optimal threshold algorithm under different conditions, especially under strong illumination, and the intelligent vehicle can track the path accurately and robustly.