A simple and reliable method is proposed to determine if a mutagen requires any metabolic activation for its genotoxicity. It consists of a comparison of the mutation frequencies induced in tester bacteria treated with the mutagen in the presence and absence of butyl isothiocyanate (Bu-NCS), a non-specific enzyme inhibitor. A mutagen requiring metabolic activation mutates the cells to a lesser extent in the presence of Bu-NCS, whereas direct-acting mutagens are capable of mutating the cells to the same extent in the presence or absence of this inhibitor. Bu-NCS seems to more or less inhibit the metabolizing ability of the enzymers present in the tester cells without any subsequent effects on the cells' viability or mutability after the inhibitor is removed. The tester bacteria to be used should preferably be deficient in the excision repair mechanism for induced deoxyribonucleic acid damage. When a carcinogen is mutagenic in such a tester bacterial system, this method would predict the ultimate, i.e., direct-acting, structure of the carcinogen because a common ultimate reactive species should be responsible for both mutagenicity and carcinogenicity. The validity of this method was tested by using 17 models of mutagenic carcinogens and the resuls supported the reliability of the prediction. Some limitations to the applicability of this method are discussed.