Thymidine phosphorylase (TP) is an important target enzyme for cancer chemotherapy because it is expressed at high levels in the hypoxic regions of many tumors and inhibitors of TP have been shown in animal model studies to inhibit angiogenesis and metastasis, and to promote tumor cell apoptosis. The 5-halo-6-[(2'-aminoimidazol-1'-yl)methyl]uracils (3, X = Cl, Br) are very potent inhibitors of E. coli and human TP with IC(50) values of approximately 20 nM when the enzyme concentration is approximately 40 nM. Their 4'-aminoimidazol-1'-yl analogues (4, X = Cl, Br) are >350-fold less active with IC(50) values of approximately 7 microM. The 5-unsubstituted analogues (3 and 4, X = H) were both less active than their 5-halo derivatives. Determination of pK(a) values and molecular modeling studies of these compounds in the active site of human TP was used to rationalize their activities. The finding that 3, X = Br has a poor pharmacokinetic (PK) profile in mice, coupled with the desire for tumor selectivity, led us to design prodrugs. The corresponding 2'-nitroimidazol-1'-ylmethyluracils (5, X = Cl, Br) are >1000-fold less active (IC(50) 22-24 microM) than their 2'-amino analogues and are reduced to the 2'-amino inhibitors (3, X = Cl, Br) by xanthine oxidase (XO). As XO is also highly expressed in many tumors, the 2'-nitro prodrugs have the potential to selectively deliver the potent 2'-aminoimidazol-1'-yl TP inhibitors into hypoxic solid tumors.