Abstract: The photon events acquired by the single-photon lidar have the characteristics of random distribution, which makes the laser ranging value be uncertain and reduces the ranging accuracy of the single-photon lidar. Under the circumstance that the sampling resolution was not decreased, an accumulative histogram of photon was constructed by accumulating photon events from the multiple contiguous spots. The range was determined on the basis of the time centroid of the derived target response function. As for the ICESat-2 single-photon lidar, a comprehensive evaluation approach of the range error was put forward by considering the effects of the terrestrial relief, the accumulative spot numbers and returned photons on the range error according to the root mean square error (RMSE) and mean absolute error (MAE). The random terrain on a strip of ICESat-2 in the West Valley City of American Utah was selected to validate the proposed method. The results demonstrate that the RMSE value is reduced from 114.25 cm to 63.84 cm and the MAE value is reduced from 70.97 cm to 48.52 cm. They outweigh the RMSE of 137.96 cm and MAE of 97.24 cm from the ICESat-2 products. The conclusions might provide some guidance for improving the ranging precision on the terrain.