A hyperresponsive tool to assess supramolecular catalysis, the cyclization of di-epoxides into cyclic ethers is used to elucidate the difference between pnictogen-bonding and Lewis acid catalysis systematically. For all stereoisomers, most tested catalysts follow the Baldwin rules. Brønsted acid and anion–π catalysis afford almost only Baldwin products. Lewis acids such as SbCl₃, BF₃ and BiCl₃ give the poorest selectivity, with at least 50 % Baldwin (B) products for all stereoisomers. In clear contrast, optimized pnictogen-bonding catalysts operating on the Sb(III) and the Sb(V) level give the fused anti-Baldwin (A) bicycles as the main product of trans epoxides, independent of syn or anti relation of the two epoxides. In the cis series, Sb(III) pnictogen-bonding catalysts afford BA products almost exclusively for syn diastereomers, while anti diastereomers give equal amounts of BA and AB products. Sb(V) pnictogen-bonding catalysts show similarly special trends. These unique characteristics support that pnictogen-bonding catalysis differs from Lewis acid catalysis and can arguably be defined as its non-covalent counterpart, just like hydrogen-bonding catalysis is understood and appreciated as the non-covalent counterpart of Brønsted acid catalysis. Computational studies on the origin of anti-Baldwin selectivity reveal Sb(V) catalysts with an introvert deep σ hole surrounded by an almost planar ring of ligands. Central pnictogen-bond attraction against peripheral steric repulsion then forces the epoxide to break open. In contrast, transient antimony oxidation in cyclic intermediates accounts for the chemoselectivity of Sb(III) catalysts. Presumably due to insufficient accessibility of their σ holes, the activity of tetrel-bonding catalysts is negligible. The division between powerful pnictogen-bonding and irrelevant tetrel-bonding catalysts does not exist with the respective orthodox Lewis acids and thus supports that σ-hole and Lewis acid catalysis are not the same.