Human alpha 2-antiplasmin (alpha 2-AP) has previously been shown to possess overlapping inhibitory sites for trypsin and chymotrypsin [Potempa, Shieh and Travis (1988) Science 241, 699-700]. Since this is currently unique among active-site-directed inhibitors of proteinases, and difficult to explain in terms of accepted inhibitory mechanisms, we re-examined the claim. Initial characterization of purified alpha 2-AP revealed an additional 12 residues preceding the published N-terminus, prompting us to revise the previous numbering. We found that trypsin caused cleavage of the Arg376-Met377 bond in the reactive-site loop of the inhibitor, whereas chymotrypsin caused cleavage at two sites in approx. equal amounts at 37 degrees C: Met374-Ser375 (site 1) and Met377-Ser378 (site 2). At 0 degrees C alpha 2-AP became a more efficient inhibitor of chymotrypsin, and the proportion of cleavage at site 1 declined, indicating that chymotrypsin prefers to react with site 2 at 0 degrees C. Inhibitors of the alpha 2-AP type are inactivated when cleaved in their reactive-site loops by proteinases that they do not inhibit, so we conclude that site 1 is treated as a substrate by chymotrypsin. Site 2 is the inhibitory site for chymotrypsin. We confirm that alpha 2-AP does indeed have overlapping reactive sites for trypsin and chymotrypsin, and since the locations of chymotrypsin-interaction sites vary with temperature, we suggest that alpha 2-AP cannot have rigid reactive-site geometry. More likely, it has a mobile reactive-site loop of the type that has been recently demonstrated for eglin C.