A large number of aziridinyl quinones represented by series 1-9 were studied with respect to their DT-diaphorase substrate activity, DNA reductive alkylation, cytostatic/cytotoxic activity, and in vivo activity. As a result, generalizations have been made with respect with respect to the following: DT-diaphorase substrate design, DT-diaphorase-cytotoxicity quantitative structure-activity relationship (QSAR), and DNA reductive alkylating agent design. A saturating relationship exists between the substrate specificity for human recombinant DT-diaphorase and the cytotoxicity in the human H460 non-small-cell lung cancer cell line. The interpretation of this relationship is that reductive activation is no longer rate-limiting for substrates with high DT-diaphorase substrate specificities. High DT-diaphorase substrate specificity is not desirable in the indole and cylopent[b]indole systems because of the result is the loss of cancer selectivity along with increased toxicity. We conclude that aziridinyl quinones of this type should possess a substrate specificity (Vmax/KM) < 10 × 10-4 s-1 for DT-diaphorase in order not to be too toxic or nonselective. While some DNA alkylation was required for cytostatic and cytotoxic activity by series 1-9, too much alkylation results in loss of cancer selectivity as well as increased in vivo toxicity. Indeed, the most lethal compounds are the indole systems with a leaving group in the 3α-position (like the antitumor agent EO9). We conclude that relatively poor DNA alkylating agents (according to our assay) show the lowest toxicity with the highest antitumor activity.