Directed enzyme prodrug therapy is a form of cancer chemotherapy in which bacterial prodrug-activating enzymes, or their encoding genes, are directed to the tumour before administration of a prodrug. The prodrug can then be activated into a toxic drug at the tumour site, reducing off-target effects. The bacterial nitroreductases are a class of enzymes used in this therapeutic approach and although very promising, the low turnover rate of prodrug by the most studied nitroreductase enzyme, NfnB from Escherichia coli (NfnB-Ec), is a major limit to this technology. There is a continual search for enzymes with greater efficiency, and as part of the search for more efficient bacterial nitroreductase enzymes, two novel enzymes from Bacillus cereus (strain ATCC 14579) have been identified and shown to reduce the CB1954 (5-(aziridin-1-yl)-2,4-dinitrobenzamide) prodrug to its respective 2′-and 4′-hydroxylamine products. Both enzymes shared features characteristic of the nitro-FMN-reductase superfamily including non-covalently associated FMN, requirement for the NAD(P)H cofactor, homodimeric, could be inhibited by Dicoumarol (3,3′-methylenebis(4-hydroxy-2H-chromen-2-one)), and displayed ping pong bi bi kinetics. Based on the biochemical characteristics and nucleotide alignment with other nitroreductase enzymes, one enzyme was named YdgI-Bc and the other YfkO-Bc. Both B. cereus enzymes had greater turnover for the CB1954 prodrug compared with NfnB-Ec, and in the presence of added NADPH cofactor, YfkO-Bc had superior cell killing ability, and produced mainly the 4′-hydroxylamine product at low prodrug concentration. The YfkO-Bc was identified as a promising candidate for future enzyme prodrug therapy.