The reactive site loop of serpins undoubtedly defines in part their ability to inhibit a particular enzyme. Exchanges in the reactive loop of serpins might reassign the targets and modify the serpin-protease interaction kinetics. Based on this concept, we have developed a procedure to change the specificity of known serpins. First, reactive loops are very good substrates for the target enzymes. Therefore, we have used the phage-display technology to select from a pentapeptide phage library the best substrates for the human prostate kallikrein hK2 [Cloutier, S.M., Chagas, J.R., Mach, J.P., Gygi, C.M., Leisinger, H.J. & Deperthes, D. (2002) Eur. J. Biochem. 269, 2747-2754]. Selected substrates were then transplanted into the reactive site loop of alpha1-antichymotrypsin to generate new variants of this serpin, able to inhibit the serine protease. Thus, we have developed some highly specific alpha1-antichymotrypsin variants toward human kallikrein 2 which also show high reactivity. These inhibitors might be useful to help elucidate the importance of hK2 in prostate cancer progression.