A statistical microphysical model of the dust, emitted into the atmosphere by cement enterprises, is developed, which allows for possible variations in the chemical composition of particles, their concentra tions, and size distribution function. An ensemble of microphysical parameters of dust was simulated and the extinction coefficient was calculated at wavelengths of 0.355, 0.532, 1.064, 1.25, 1.56, 1.67, and 2.14 μm from the "windows of transparency" of exhaust gas at cement enterprises. The errors of retrieval of the mass con centration of dust from the data of optical sensing at wavelengths of a Nd:YAG laser with third harmonic gen eration, are estimated on the basis of a multiple regression method. The sets of two (0.532 and 2.14 μm) and three (0.532, 1.56, and 2.14 μm) wavelengths of optical sensing are ascertained, which are optimal in terms of their information content about the concentration and the resistance of the corresponding solutions of the inverse problem to measurement errors, and the effect of dust particle shape. Dependencies of errors of con centration retrieval, from the extinction coefficient at optimal wavelengths, on the optical measurement error are calculated. A tight correlation between the dust concentration and its extinction coefficient at a wave length of 2.14 μm was derived; it allows retrieval of the concentration from fixed frequency optical sensing data with an error of about 8%.