Abstract In the assay of two soybean trypsin inhibitors, the Kunitz and the Bowman-Birk inhibitors, two procedures were used: the current procedure in which the substrate is added last (the S-last test), after inhibitor is mixed with enzyme, and a new procedure in which the enzyme is added last (the E-last test), after inhibitor is mixed with substrate. In the E-last test, the inhibition values obtained were independent of the premix pH and preincubation time, while in the S-last test, the values were functions of these two parameters. When the pH was below 2.7 or near neutrality, the values of S-last test were equal to those of the E-last test. When the pH was 2.7 – 5.5 or 7.5 – 9.0, the S-last values were lower than the E-last values. This so-called “reactant sequence effect” is attributed to limited hydrolysis of the inhibitor at these pH ranges, in accordance with the reactive site model proposed by K. Ozawa and M. Laskowski, Jr. (1966, J. Biol. Chem. 241, 3955). When the premix pH was jumped from the acidic or alkaline ranges to near neutral, the reactant sequence effect was abolished, indicating resynthesis of the inhibitor from the modified one. Results of this study show that the E-last test is preferable to the S-last test for assaying a trypsin inhibitor of protein nature.