The current therapies for prostate cancer involve either surgical resection or drugs which produce androgen deprivation. These methods, however, induce shrinkage of hyperplastic tissue via the blockade of the action of testosterone, and therefore cause undesirable side-effects such as gynaecomastia and impotence. The enzyme 5α-reductase has been under investigation as a possible chemotherapeutic target in the fight against prostate cancer, since it is the enzyme responsible for the conversion of testosterone to the more potent androgen, dihydrotestosterone. In the present study, we have modelled both steroidal and non-steroidal inhibitors of 5α-reductase in an attempt to elucidate the essential structural requirements needed in the design of non-steroidal inhibitors. The study suggests that there is a requirement for groups to mimic the steroid substrate A-ring, in particular the C(3) carbonyl group; the area about C(3), C(4), C(5) and C(6) of testosterone appears to be sterically hindered, presumably due to the binding of the NADPH moiety; and the area of the active site about the C(17) of the steroid substrate does not appear to possess hydrogen bonding groups and is not restricted. The study also suggests that the two isozyme types possibly vary in the positioning of the reducing NADPH moiety, i.e the preference for the 4- and 6-azasteroids is dependent on the positioning of the NADPH about the steroid backbone. Using data obtained from this study, several non-steroidal inhibitors have been designed, synthesized and subsequently tested-some of which have been found to possess good inhibitory activity.