Most starch hydrolases and related enzymes belong to the e-amylase family which contains a characteristic catalytic (fi/e)8-barrel domain. Currently known primary structures that have sequence similarities represent 18 different specificities, including starch branching enzyme. Crystal structures have been reported in three of these enzyme classes: the e-amylases, the cyclodextrin glucanotransferases, and the oligo-l,6-glucosidases. Throughout the e-amylase family, only eight amino acid residues are invariant, seven at the active site and a glycine in a short turn. However, comparison of three-dimensional models with a multiple sequence alignment suggests that the diversity in specificity arises by variation in substrate binding at the fl--+ e loops. Designed mutations thus have enhanced transferase activity and altered the oligosaccharide product patterns of e-amylases, changed the distribution of e-,/3-and 7-cyclodextrin production by cyclodextrin glucanotransferases, and shifted the relative e-1,4: e-1,6 dual-bond specificity of neopullulanase. Barley e-amylase isozyme hybrids and Bacillus e-amylases demonstrate the impact of a small domain B protruding from the (/3/e)8-scaffold on the function and stability. Prospects for rational engineering in this family include important members of plant origin, such as e-amylase, starch branching and debranching enzymes, and amylomaltase.Abbreviations: CGTase, cyclodextrin glucanotransferase; SBD, starch binding domain; TAA, takaamylase A; TIM, triose-phosphate isomerase. The mutations are described with the one-letter code, i.e.