Failure of current chemotherapeutic agents to effectively treat human brain tumors has prompted the search for alternative regimens based on the inherent metabolic pathways of target cells. One way to accomplish this goal would be to design drugs in an inactive form, which upon entry into the cell would be transformed to a toxic metabolite by a naturally occurring pathway. One such pathway may be the reductive activation of naphthoquinones with one or two side chains capable of alkylation, such as 2,3-dibromomethyl-1,4-naphthoquinone (DBNQ). This reductive activation can be catalyzed by the flavoprotein DT-diaphorase [NAD(P)H:quinone oxidoreductase]. We have found that both rat 9L and some human brain-tumor cell lines contain very high levels of this enzyme and that halogenated dimethyl naphthoquinones, such as DBNQ, are highly toxic to these cells in vitro. Moreover, we have found that the cytotoxic effects of DBNQ on human tumor and murine bone marrow stem cells can be prevented or lessened by pretreatment of the cells with dicoumarol, a potent inhibitor of DT-diaphorase. Since dicoumarol does not cross the blood-brain barrier, the potential exists for human brain tumors to be destroyed with halogenated dimethylquinones and for peripheral host toxicity to be prevented by coadministration of dicoumarol.