Statistical optical aerosol models, one of which is global and several are regional, are formulated for the tropospheric layer 0-12 km. Examples of using these models to solve different problems are given. As the basis for modeling, the tropospheric-aerosol optical model OPAC formulated by using experimental data on ten aerosol fractions, including water-soluble, water-insoluble, soot, and mineral particles, sea salt, etc., is taken. Statistical characteristics of the constructed ensembles are analyzed for the spectral region 0.25-2.0 μm and scattering angles 0°-180°. It is shown that the model statistics correlate well with local data of independent measurements. The eigenvectors of the covariance matrices of the aerosol extinction coefficients (AECs) are used for optimal parametrization of the AEC spectral dependence. It is found that the optimal parametrization has a significant advantage over the parametrization based on the Ångström formula. The errors in the numerical simulation of regression retrieval of the aerosol 90°-scattering coefficient from the AEC data of the SAGE-II and SAGE-III satellite experiments are analyzed. It is shown that the regression errors range between 18 and 80% at an a priori variation in the tropospheric aerosol scattering coefficient greater than 300%. The linear-regression parameters of the aerosol backscattering coefficient are found from AEC data for three (global, sea, and desert) atmospheric models and a large number of wavelengths that can be used in solving the lidar equation as applied to atmospheric sounding.