Recent investigations have indicated that one way in which to achieve selectively toxic antimetabolites is to find a biological precursor of a vitamin, and then to construct a suitable analog of it. Such an analog may then harm those species which synthesize this vitamin, and should not affect those which have a nutritional requirement for it. Thus, the selective toxicity of sulfanilamide and its congeners for certain bacteria in contrast to higher animals has been postulated to reside in the difference in requirement for folic acid (1-3). This postulate has received support in the finding that the harmful action of a pimelic acid analog could be foretold with fair precision from a knowledge of nutritional requirements for biotin (4). The vitamin which arises from the precursor by a series of biosynthetic reactions (e.g., biotin) sometimes may be able completely and non-competitively to erase the effects of the analog.The present investigation was undertaken with two questions in mind. (a) Could additional substances of predictable selective toxicity be realized by construction of structural analogs of a precursor of another vitamin? (b) Could a means be found to foretell antimetabolites of the precursor which would not be counteracted by the vitamin?With regard to the first question, the correlation of the selective toxicity of sulfanilamide and of the pimelic acid analog with nutritional requirements for the vitamins, folic acid or biotin, was just cited. In the present study a single precursor of both riboflavin and vitamin B~2 was sought, an analog of this precursor was prepared, and it has been shown to retard the growth of all species examined except those with a nutritional need for these two vitamins. This demonstration added evidence in favor of the postulate that selectively toxic substances can be achieved in this general way.The bearing of the second question on some of the problems of chemotherapy and of pharmacology can be seen from consideration of the action of sulfanilamide, in contrast to that of the pimelic acid analog mentioned above. This