This paper presents a CCD based stereovision algorithm (called the ”v-disparity” algorithm [1]) for onboard road obstacles detection (pedestrians, vehicles, motorbikes, etc.) in night condition. The algorithm is explained, and then evaluated towards different obstacles. The theoretical good properties of the ”v-disparity” algorithm - accuracy, robustness, computational speed - are experimentally confirmed. Experiments show that obstacles are detected in a precise manner with high confidence values, at frame rate (25 Hz) using no special hardware.