The discovery of therapeutic agents has, in the past, started primarily with random screening efforts. These screens, although effective, are time-consuming and expensive. Attempts are now being made to design more efficient methods of screening that take into account available information about the three-dimensional structure of a target receptor or enzyme. In the case of acquired immunodeficiency syndrome, the structure of a proteolytic enzyme, the human immunodeficiency virus (HIV) protease, an aspartyl protease that plays a crucial role in the viral life cycle, has been determined and extensively characterized. Using the protease structure and the computer program DOCK, the active site of the protease was mapped and its shape used to screen a subset of the Cambridge Structural Database. Among the molecules whose shape was complementary to the active site was the antipsychotic agent haloperidol. This molecule and several chemically modified derivatives were shown to bind competitively with micromolar affinity to the HIV protease but not to cellular aspartyl proteases. X-ray structures of the HIV protease complexed with haloperidol derivatives show the molecules binding in the predicted position at the active site. In an attempt to overcome the problems associated with low-affinity competitive inhibitors, reactive groups that enable the molecule to serve as a specific irreversible inhibitor of the HIV protease were introduced onto the haloperidol scaffold. These inhibitors demonstrated an ability to block viral polyprotein processing in a tissue culture model of HIV-1 infection, although their cytotoxicity is pronounced.(ABSTRACT TRUNCATED AT 250 WORDS)