Activities of enzymes involved in the metabolic formation and catabolism of epoxides were determined in liver subcellular preparations from 11 mammalian species and various strains of mice. The most conspicuous finding was that the activities of the microsomal epoxide hydrolase were clearly lower in the mouse than in the other species. This invited the working hypothesis that epoxides may be involved in mouse liver carcinogenesis. The carcinogens may be metabolised themselves to reactive epoxides or they may modify the metabolism of epoxides formed from endogenous or other foreign compounds. To examine the former point, phenobarbital, DDT (1,1-bis(p-chlorophenyl)-2,2,2-trichloroethane), lindane and benzo(a)pyrene were investigated for mutagenicity in Salmonella typhimurium using as the carcinogen-metabolising system subcellular liver preparations from animals in which these compounds efficiently induce liver tumours and from resistant animals. Phenobarbital, DDT and lindane were not mutagenic under any conditions, including those where microsomal epoxide hydrolase was also inhibited. However, a DDT metabolite, 1,1-bis(p-chlorophenyl)-2,2-dichloroethane was mutagenic in strain TA98, when norharman was added to the metabolising system, rat liver postmitochondrial fraction. Benzo(a)pyrene, which efficiently induces liver tumours in male but not in female newborn C3HeB/FeJ X A/J mice, was similarly activated by liver preparations from male and female animals. This was true with and without pretreatment of the mice with an inducer of cytochrome P-448. Also, activities and inducibilities of monooxygenase, epoxide hydrolase and glutathione transferase (toward benzo(a)pyrene and benzo(a)pyrene 4,5-oxide, respectively) were indistinguishable between males and females. Therefore, differences in the metabolism of benzo(a)pyrene do not appear to be the reason for the sex difference in tumour susceptibility. Likewise, mouse strains with high and low frequencies of spontaneous and chemically-induced liver tumours did not appreciably differ in their hepatic microsomal epoxide hydrolase activities. The low level of this activity therefore cannot constitute the critical factor for the high tumour susceptibility of certain strains of mice. However the statement does not preclude potentiation of the susceptibility toward particular carcinogens owing to this metabolic trait of the mouse.