Due to the strong intensity of lidar return signal in the near field, the photon-counting system in lidar applications suffers from the pile-up effect, which can be corrected by the parameter of dead time. A mathematical model is constructed to calculate the spatial variance, which is used to evaluate the quality of Poisson distribution of lidar data. Furthermore, the analysis results of the spatial variance are utilized to estimate the dead time of the lidar photon-counting system and to correct the pile-up effect. The calculation results indicate that the photon-counting data in the far field of lidar conforms to the Poisson distribution, but the data in the near field does not. However, the Poisson distribution characteristic of the photon-counting data can be improved notably by the dead time correction. Therefore, the system dead time can be estimated by minimizing the deviation between the variance and the mean of lidar data. One can correct the pile-up effect of photon-counting data and make the data profile conform to the Poisson distribution at the maximum degree. The results show that the dead time of the photon-counting system in Licel transient recorder TR40-160 is approximately 3.402 ns, which is estimated by using the photon-counting data in the application of long-range scanning lidar. The pile-up effect in the presented photon-counting data is improved obviously.