Medium effects are normally studied by comparing the rates of reactions in different solvents. However, medium effects at the active site of enzymes differ dramatically from bulk solvents, both in their diversity (the presence of more than one type of "solvent") and in their spatial arrangement. We describe medium effects in a simple catalytic system, obtained by systematic alkylation of a polymeric scaffold bearing amine groups to give synzymes that catalyze the Kemp elimination of benzisoxazoles with remarkable efficiency. Our analysis indicates that catalysis by these synzymes is driven primarily by specific, localized enzyme-like medium effects, and these effects seem to differ dramatically from the nonspecific medium effects (i.e., desolvation activation) exhibited by solvents. Ligand-binding studies indicate that the synzyme active sites provide localized microenvironments affording a combination of hydrophobic and apolar regions on one hand and dipolar, protic, and positively charged on the other. Such localized microenvironments are not available in bulk solvents. A Brønsted (leaving group) analysis indicates that, in comparison to solvent catalysis, the efficiency of synzyme catalysis shows little sensitivity to leaving group pK(a). We show that enzyme-like medium effects alone, in the absence of efficient positioning of the catalytic amine base relative to the substrate, can give rise to rate accelerations as high as 10(5), for both activated and nonactivated substrates. Supported by the accidental identification of active sites on the surfaces of noncatalytic proteins and the promiscuous activities found in many enzymes, our findings suggest that the interfaces of protein surfaces and their hydrophobic cores provide a microenvironment that is intrinsically active and may serve as a basis for further evolutionary improvements to give proficient and selective enzymes.