Recent experimental work with steels of carefully controlled purity has allowed the phenomenon of temper brittleness to be defined more closely. It is now seen as a reversible embrittlement to which alloy steels of commercial purity are subject if exposed for prolonged periods in the temperature range 400 to 600 C. It does not occur in alloy steels synthesized from high-purity elements; neither does it occur in carbon steels even of commercial purity. A detailed consideration of the equilibrium segregation theory of grain-boundary embrittlement, as well as some associated mechanisms advanced in recent years, has shown them all to be inadequate as explanations of the peculiar conditions under which temper-brittleness can arise. A modified theory of “double segregation” is proposed to explain the influence of major alloying elements on the incidence of embrittlement by impurity elements. In this theory, grain-boundary enrichment with alloying elements such as manganese, chromium, and molybdenum during austenitizing can lead to enhanced segregation of the embrittlement elements such as phosphorus, arsenic, antimony, or tin by chemical interaction. Only the latter elements cause the shifts in transition temperature and the changeover from cleavage to intergranular brittle fracture which are characteristic of temper brittleness. They do this by lowering the intergranular fracture energy γ1. On the basis of the theory the specific nature of the temper-brittleness phenomenon, as well as the relative influences of individual major alloying elements, can be explained.