Abstract Grain boundaries strongly influence the properties of polycrystalline materials, which are used for practical applications. This effect originates from the energy difference between the internal interfaces and the crystal volume, and consequently, from an interaction of the grain boundaries with other lattice defects such as point defects and dislocations, resulting in substantial reduction of the total Gibbs free energy of the system. One representative of such interaction – the interaction with the solute atoms (nanosegregation at the grain boundaries) – is of high importance because it predominantly affects the intergranular cohesion and, in consequence, the mechanical behavior of the materials. To improve the mechanical properties of the technically used materials, the concept of Grain Boundary Engineering was proposed in 1980s by Tadao Watanabe to produce the polycrystals with optimum properties by controlling the grain boundary character distribution. For this purpose, the knowledge of the broad spectrum of properties of the broad spectrum of grain boundaries is needed. One of the most important properties which affects the behavior of polycrystalline materials and which has thus be considered in the Grain Boundary Engineering concept, is the nanosegregation of the solutes at the grain boundaries. The chemical composition of individual grain boundaries can be either measured experimentally using various techniques of surface analysis (for example AES, ESCA and SIMS), or simulated by the methods of theoretical modeling (for example Monte Carlo and molecular dynamics). Both these approaches, however, have serious limitations, which do not allow providing us with the data for a broad spectrum of the solutes and for a broad spectrum of the grain boundaries. Therefore, any prediction of such data seems to be very prospective and desirable. A detail analysis of the effect of the solid solubility on variations of the Gibbs free energy of interfacial segregation resulted in formulation of two simple expressions. One of them is the so-called grain boundary segregation diagram, relating the standard molar enthalpy of solute segregation, ?HI0, to the solid solubility data, T??lnXI?(T?), and reflecting anisotropy of grain boundary segregation, ?HI?(?, XI?=1). The other expression, so-called the enthalpy-entropy compensation effect shows a close relationship between the standard molar enthalpy, ?HI0, and the standard molar entropy, ?SI0, of grain boundary segregation, . Based on the above two relationships the segregation of any solute at any grain boundary of ?-iron can be predicted. This prediction, which offers a substantial extension of the database on grain boundary segregation is tested by comparing with the literature data on experimental measurement of grain boundary segregation in polycrystalline materials of binary, pseudobinary or multicomponent systems. In addition, the potency of the prediction method is shown to display chemical composition of the grain boundaries in two technically used materials, the low-alloy ferrite steel and the ferrite phase of the nodular cast iron. Due to the generality of the derivation of the above relationships, the proposed prediction method can be simply extended to different matrices and types of the interfaces (free surfaces, phase interfaces).