A surrounding monitoring apparatus obtains the position of a pedestrian contained in an image captured by a camera mounted on a vehicle on the basis of the position of the pedestrian in the image, and determines, on the basis of the obtained position, whether or not the possibility of collision with the pedestrian is high. In the apparatus, an error in extracting the pedestrian position may increase temporarily. In view of this, when the magnitude of change in a position correlation value representing the position of the pedestrian per a predetermined time (position change amount) exceeds a change amount upper limit value, the apparatus modifies the position correlation value such that the position change amount becomes equal to the change amount upper limit value and determines, on the basis of the modified position correlation value, whether or not the possibility of collision with the pedestrian is high.