A mathematical model establishing relations between the selectivity, degree of ion exchange and structure of spatial polyelectrolytes has been developed. The structure is expressed through two parameters: radius of interaction (that is a maximal distance at which the exchange sites influence each other) and the number of exchange sites situated in the sphere of interaction (neighbors). A larger number of neighbors should result in more complicated shape of property-composition dependencies of the ion exchange system. A computer model allowing estimation of the probability of the presence of different numbers of neighbors in different elements of the polymer network as a function of distance from the exchange site was suggested. The polymer network is represented as a body obtained by integration of fragments with eight to ten monomeric units of different shapes: linear chains, H and T shape units and small rings. The theory was applied to interpreting dependencies of selectivity of cation exchange on sulfonic styrene-divinylbenzene cation exchangers and strong base anion exchangers on the same matrix. Dependencies of selectivity on degree of ion exchange and cross-linkage of the resins are in good agreement with the prediction of the theory if the predominant number of neighbors in different particular cases is ranging between two and four. The non-equivalence of exchange sites of spatial polyelectrolytes is their inherent property because different exchange sites have different amounts of the neighbors. The difference in the selectivity dependence on the cross-linkage of cation and anion exchangers on styrene-divinylbenzene matrix finds a simple and natural explanation. (C) 2002 Elsevier Science B.V. All rights reserved.