Simulation models are often used in “soft monitoring systems” to predict effects of production and to diagnose the technological process weather it meets required standards during the analyzed period. They have potential to be used in mineral and coal processing plants where information on process performance is gained from time and labor consuming laboratory tests on samples of material taken from the feed and products from the beneficiation process. Coal concentrates from raw coal are produced through gravitational separation processes in heavy media vessels and in jigs. Both processes are described by a similar model of raw coal washability characteristics and a partition curve of the machine. The concept of a soft monitoring for on-line identification of the partition curve (imperfection) and the washability characteristics is based on comparing continuously signals from belt scales (yields of three products) and two separation densities rom radiometric density meters, at two different pairs of separation densities in a 3-product gravitational process in a jig, with similar signals generated in a simulation model of the process. The simulation analysis has shown that the presented soft monitoring system can be applied efficiently for a complex identification of a coal separation process in a gravitational washer such as a jig. On-line determination of the raw coal washability characteristics, the partition characteristics of the machine and separation densities can be evaluated in this way. Good results for three density fractions have been obtained. Results of imperfection soft monitoring show that it is possible to classify on-line the efficiency of a jig into several ranges, for example: very good, good, satisfactory, not satisfactory.