A robust detection of obstacles and other vehicles on the path of a moving vehicle is essential for a number of the basic guidance functionalities, such as Road Following or Platooning. Therefore a critical analysis of the performance of Obstacle Detection is mandatory. This work presents a critical analysis of a stereo vision-based guidance system installed onto the ARGO autonomous vehicle. Different obstacles -with various shape and size- have been placed in front of the vision system in a number of different positions. The results obtained by the guidance system have been collected and analyzed. This paper surveys the set of results obtained so far, highlights their characteristics, and discusses the main advantages and problems of such a solution.