To solve the short-range precise navigation problem in the probe-and-drogue autonomous aerial refueling, a binocular vision-based short-range navigation method was presented. This method used two cameras installed on the receiver to capture the images of the optical markers mounted on the end surface of the drogue, then detected the light spot regions in the images by contour tracking algorithm and calculated their gray scale centers as the image points of the optical markers. These image points were matched according to epipolar geometry constraints, and the position and attitude of the drogue were calculated using space circle fitting arithmetic. A three-dimensional virtual scene of probe-and-drogue aerial refueling was established to simulate this navigation method. Simulation results show that this method has a high position and attitude measurement accuracy during the short-range docking maneuver, fulfilling the demand of autonomous aerial refueling.