One of the most dangerous technical failures of materials is intergranular brittle fracture ( temper embrittlement) as it proceeds very quickly and its appearance is often hardly predictable. It is known that this phenomenon is closely related to the chemistry of grain boundaries and to the difference of the segregation energies of the grain boundaries and the free surfaces (RiceWang model). To elucidate the effect of individual solutes on embrittlement of various materials such as steels and nickel-base superalloys, grain boundary and surface segregation was extensively studied in many laboratories. As a result, numerous data on surface and grain boundary segregation have been gathered in literature. They were obtained in two main ways, by computer simulations and from experiments. Consequently, these results are frequently applied to quantify the embrittling potency of individual solutes. Unfortunately, the values of the segregation energy of a solute at grain boundaries as well as at the surfaces obtained by various authors sometimes differ by more than one order of magnitude: such a difference is unacceptable as it cannot provide us with representative view on the problem of material temper embrittlement. In some cases it seems that these values do not properly reflect physical reality or are incorrectly interpreted. Due to the above mentioned large scatter of the segregation and embrittlement data a critical assessment of the literature results is highly needed which would enable the reader to avoid both the well known and less well known pitfalls in this field. Here we summarize the available data on interfacial segregation and embrittlement of various solutes in nickel and bcc iron and critically discuss their reliability, assessing also limitations of individual approaches employed to determine the values of segregation and strengthening/embrittling energies, such as density functional theory, Monte Carlo method, molecular statics and dynamics and tight binding on the theoretical side, and Auger electron spectroscopy, 3D tomographic atom probe, and electron microscopy techniques on the experimental side. We show that experimental methods have serious limitations which can be overcome by accepting reasonable assumptions and models. On the other hand, the theoretical approaches are limited by the size of the computational repeat cell used for the calculations of the segregation energy. In both cases, a careful critical analysis of the available segregation energy and/or enthalpy reflecting physical reality allows to assess the reliability of these values and their applicability in analysis of intergranular brittle fracture in steels and nickel-base alloys. (C) 2016 Elsevier Ltd. All rights reserved.