The carcinogenicities for rats and mice of N-methyl-4-aminoazobenzene (MAB) and its hepatic microsomal metabolite N-hydroxy-N-methyl-4-aminoazobenzene (N-hydroxy-MAB) were compared under several conditions. N-Ethyl-4-aminoazobenzene, 4-aminoazobenzene, and their N-hydroxy derivatives were also included in some of the assays. About 25% of the rats given MAB or N-hydroxy-MAB (3 to 5 mmol/kg body weight) by stomach tube over a 5-week period developed hepatic tumors by 18 to 22 months. Similarly treated rats subsequently given phenobarbital in the drinking water until the termination of the experiment developed about twice as many hepatic tumors. N-Hydroxy-MAB, administered p.o., but not MAB, also induced multiple papillomas and extensive carcinomas of the forestomach in approximately 50% of the rats. Only low incidence of hepatocellular carcinomas occurred in partially hepatectomized rats given a single i.p. injection of 180 mumol/kg body weight of MAB or N-hydroxy-MAB with or without subsequent administration of phenobarbital. Although repeated s.c. doses of N-benzoyloxy-N-methyl-4-aminoazobenzene induced sarcomas at the injection site in 90% of the rats, only 3 of 20 rats developed sarcomas at the site of s.c. injections of N-hydroxy-MAB. N-Ethyl-4-aminoazobenzene, 4-aminoazobenzene, and their N-hydroxy derivatives did not induce significant numbers of tumors in any of the above assay systems. Administration to preweaning male mice of MAB, N-hydroxy-MAB, N-hydroxy-N-ethyl-4-aminoazobenzene, and N-hydroxy-4-aminoazobenzene resulted in high incidences and high multiplicities of hepatic tumors (averages of 5 to 7 tumors/mouse) within 1 year. N-Ethyl-4-aminoazobenzene and 4-aminoazobenzene also induced hepatic tumors under the same conditions, but they were less active. These data support the conclusion that the N-hydroxy metabolites of these aminoazo dyes are proximate carcinogens.