The paper presents a multidimensional analysis of mineral processing feeds consisting of different amounts of different size and density fractions. The considered feed was coal which was screened into size fractions which were subsequently separated into density fractions and their weights determined. The feed material was characterized with commonly used size and density frequency and cumulative distribution plots and next approximated with the Weibull (size) and logistic (density) mathe- matical functions. Having the contribution of each particle size and density fraction in the feed a two– dimensional analysis of the feed size/density properties was performed using two methods. The first one is based on the best chosen cumulative frequency function for two random variables and the second uses the so–called Morgenstern family functions. In the paper the undependability of the particles size and density was investigated using statistical approach based on the so–called chi² test, and the correlation between these parameters using the so–called F–Snedecor statistical test. In both cases it was found that particles size and density of the investigated coal particles were dependent what means that with growth of particle size its density grew too and there was correlation between them regardless of significance level assumed for the analysis.