The inhibition of bovine pancreatic trypsin was studied at pH 7, 25 degrees C, using mixtures of purified human alpha 2-macroglobulin (alpha 2M) and alpha 1-proteinase inhibitor (alpha 1 PI). The partitioning of the enzyme between the two inhibitors was determined by comparing control esterase activity, assayed with N-benzoyl-L-arginine ethyl ester as substrate, with that remaining after incubation with inhibitory mixtures. (At [I]0 > [E]0, remaining esteratic activity reflects the concentration of alpha 2M-associated enzyme (alpha 2M-E*) and the concentration of alpha 1PI-associated, inactive enzyme (alpha 1PI-E*) is given by the difference, [E]0-[alpha 2M-E*].) The pattern of product distribution was found to be incompatible with an inhibitory model involving parallel, second-order reactions of E with alpha 2M and alpha 1PI. The data pointed to complex formation between the two inhibitors, limiting the level of alpha 2M readily available for reaction with E. Analysis based on the binding equilibrium, alpha 2M (dimeric unit) + alpha 1PI reversible alpha 2M-alpha 1PI, yielded Kd = 2.1 +/- 0.3 microM. Complex formation between alpha 2M and alpha 1PI was verified by gel permeation experiments. alpha 2M was found to restrict the volume of distribution of alpha 1PI in Sephadex G200 beds. Kd, deduced from gel permeation behaviour, was 0.8 +/- 0.32 microM. Preliminary kinetic experiments with dialyzed plasma suggested that the alpha 2M-alpha 1PI interaction is effective also in vivo. Given Kd and the mean plasma levels of the two inhibitors ([alpha 2M] = 2 microM; [alpha 1PI] = 36 microM), it was estimated that > 90% of alpha 2M in human circulation must be complexed to alpha 1PI and lack immediate antiproteinase activity.