CB 1954 [5-(aziridin-1-yl)-2,4-dinitrobenzamide] has been the subject of continued interest for over 30 years. As an anti-cancer agent, it represents one of the very few examples of a compound that shows real anti-tumor selectivity. Unfortunately, for the treatment of human disease, this anti-tumor selectivity was seen only in certain rat tumors. The basis for the anti-tumor selectivity of CB 1954 is that it is a prodrug that is enzymatically activated to generate a difunctional agent, which can form DNA-DNA interstrand crosslinks. The bioactivation of CB 1954 in rat cells involves the aerobic reduction of its 4-nitro group to a 4-hydroxylamine by the enzyme NQO1 (DT-diaphorase). The human form of NQO1 metabolizes CB 1954 much less efficiently than rat NQO1. Thus human tumors are insensitive to CB 1954. In view of the proven success of CB 1954 in the rat system, it would be highly desirable to re-create its anti-tumor activity in man. This has led to the development of CB 1954 analogs and other prodrugs activated by nitroreduction such, as those based on a self-immolative activation mechanism. A gene therapy-based approach for targeting cancer cells and making them sensitive to CB 1954 and related compounds has been developed. VDEPT (gene-directed enzyme prodrug therapy) has been used to express an E. coli nitroreductase in tumor cells and human tumor cells transduced to express this enzyme are very sensitive to prodrugs activated by nitroreduction. CB 1954 is in clinical trial for this application. Recently it has been shown that a latent nitroreductase is present in some human tumors. This is NQO2 - an enzyme that requires for activity, the non-biogenic compound dihydronicotinamide riboside (NRH) as a cosubstrate. When active, NQO2 is 3000 times more effective than human DT-diaphorase in the reduction of CB 1954. NRH and reduced pyridinium derivatives that, like NRH, act as co-substrates for NQO2, produce a dramatic increase in the cytotoxicity of CB 1954 against human cell lines in vitro and its anti-tumor activity against certain human xenografts in vivo. NQO2 activity is substantially raised in tumor samples from colorectal and hepatoma patients (up to 14-fold). A phase I clinical trial of an NQO2 co-substrate with CB 1954 is scheduled.