Atmospheric aerosol particles influence the Earth’s radiation balance both directly and indirectly. The aerosol size distribution (ASD) is one of the most important microphysical properties. In this paper, the generalized cross-validation (GCV) regularization method is used for the retrieval of ASD from three-wavelength lidar optical data. The numerical simulations are carried out using synthetic backscatter and extinction coefficients. Simulations results demonstrate that the ASD depends on particle refractive index. Choosing the suitable refractive index is crucial to retrieve aerosol size distribution accurately. Moreover, the numerical results show that, for the same refractive index, it is more suitable to retrieve broad ASD, which has larger mode width σ. The GCV regularization method has been tested for a set of experimental data from three-wavelength lidar, which provides backscatter coefficient at 355, 532 and 1064 nm and extinction coefficient at 355 and 532 nm. Experimental result shows that the retrieved size distribution belongs to the urban industrial type and fine mode. The result shows good agreement with the actural atmospheric aerosol size distribution of local area. Both the simulation and the expriment demonstrate that the GCV regularization method is feasible to retrieve the aerosol size distribution.