N-(Benzyloxycarbonyl)-L-aspartyl-L-phenylalanine methyl ester (Z-AspPheOMe), a precursor of the synthetic sweetener aspartame, was synthesized, using thermolysin immobilized onto Amberlite XAD-7, both in ethyl acetate and in tert-amyl alcohol. The initial rates for synthesis of Z-AspPheOMe in the organic solvents were predicted on the basis of a model proposed for an aqueous/organic biphasic reaction and compared with the experimentally observed substrate concentration dependencies. The experimental synthetic rates using the enzyme immobilized at a high enzyme concentration were lower than the calculated ones over a wide range of the substrate concentration. It was suggested as a reason for this discrepancy that the enzyme molecules form compact aggregates and those existing inside the aggregates cannot be utilized for reaction. The experimental results with the enzyme immobilized at a low concentration in ethyl acetate coincided well with the calculated ones. On the other hand, when tert-amyl alcohol was used, the experimental results were different in tendency irrespective of the amount of enzyme loaded, probably due to the fact that a distinct water phase does not exist around the enzyme aggregates inside the support.