N-Acetylneuraminic acid lyase (NAL) exhibits poor facial selectivity during carbon-carbon formation, and as such, its utility as a catalyst for use in synthetic chemistry is limited. For example, the NAL-catalyzed condensation between pyruvate and (2R,3S)-2,3-dihydroxy-4-oxo-N,N-dipropylbutyramide yields ca. 3:1 mixtures of diastereomeric products under either kinetic or thermodynamic control. Engineering the stereochemical course of NAL-catalyzed reactions could remove this limitation. We used directed evolution to create a pair of stereochemically complementary variant NALs for the synthesis of sialic acid mimetics. The E192N variant, a highly efficient catalyst for aldol reactions of (2R,3S)-2,3-dihydroxy-4-oxo-N, N-dialkylbutyramides, was chosen as a starting point. Initially, error-prone PCR identified residues in the active site of NAL that contributed to the stereochemical control of an aldolase-catalyzed reaction. Subsequently, an intense structure-guided program of saturation and site-directed mutagenesis was used to identify a complementary pair of variants, E192N/T167G and E192N/T167V/S208V, which were similar to 50-fold selective toward the cleavage of the alternative 4S- and 4R-configured condensation products, respectively. It was shown that wild-type NAL could not be used for the highly stereoselective synthesis of a 6-dipropylamide sialic acid mimetic because the 4S-configured product was only similar to 3-fold kinetically favored and only similar to 3-fold thermodynamically favored over the alternative 4R-configured product. However, the complementary 4R- and 4S-selective variants allowed the highly (> 98:< 2) diastereoselective synthesis of both 4S- and 4R-configured products under kinetic control from the same starting materials. Conversion of an essentially nonselective aldolase into a pair of complementary biocatalysts will be of enormous interest to synthetic chemists. Furthermore, since residues identified as critical for stereoselectivity are conserved among members of the NAL superfamily, the approach might be extended to the evolution of other useful biocatalysts for the stereoselective synthesis of biologically active molecules.