The question of an interfacial versus ahomogeneous reaction is carefully addressed for the enzymatic biphasic cleavage of mandelonitrile to benzaldehyde by Prunus amygdalus hydroxynitrile lyase (pa-HnI) (Hickel et al. [1999] Biotechnol Bioeng 36:425-436). Experimental evidence, including 1) the reaction ceases when the interface is populated by previously adsorbed denatured pa-HnI, 2) the reaction continues even after washout of the bulk enzyme from the aqueous phase, 3) highly nonpolar organic solvents initially promote fast reaction kinetics that relatively quickly decay to zero product production, and 4) the reaction rate is nonlinear in the bulk enzyme concentration, provide robust grounds for an interfacial reaction. We also model enzymatic mandelonitrile cleavage assuming a homogeneous aqueous-phase reaction. The homogeneous reaction scheme does not simultaneously account for the experimental observations of a linear dependence of the reaction rate on organic/water interfacial area, no dependence on the aqueous-phase volume, and a nonlinear dependence on pa-HnI aqueous concentration. Further, simple calculations demonstrate that the homogeneous reaction rate is at least three orders of magnitude slower than those observed by Hickel et al. (1999). We again conclude that enzyme adsorbed at the organic solvent/water interface primarily catalyzes the biphasic mandelonitrile cleavage reaction. (C) 2003 Wiley Periodicals, Inc.