The laboratory investigations which are necessary to evaluate the mineral processing effectiveness usually include density separation of narrow size fraction of the feed and one of the products. In each size - density fractions the content of valuable mineral has to be determined. These data and yield in each size - density fraction can be specified by a matrix of partition coefficients. They are counterparts of recoveries. These coefficients can be compared with the mean recoveries at each density or size fraction or the total recovery, to determine which density - size fractions are processed in the best or worst way. The basic categories are: vector of metal content l _i 1711, 7t2, , ltM I vector of particle size distribution x = I d1, d2, ..., dN I matrix of partition coefficients P11, P12, Pi N P _ P21, P22, ..., P2 N PM1 PM2..., PMN These three conceptions allow one to determine all the basic metallurgical parameters of a mineral processing unit. Losses of valuable mineral are connected with: mineralogical composition of the ore (types of minerals, their properties, dissemination, crystallochemical affinity, type and properties of gangue minerals), applied technology and technical equipment (beneficiation as a stochastic process, processing methods, process control, intergrowth liberation, water quality). Examples of application of the analytical method proposed are given for zinc - lead and coal processing plants.