The role of the key catalytic residues Glu134 and Glu138 in the retaining 1,3-1,4-beta-glucanase from Bacillus licheniformis is probed by a chemical rescue methodology based on enzyme activation of inactive mutants by the action of added nucleophiles. While Glu134 was proposed as the catalytic nucleophile on the basis of affinity labeling experiments, no functional proof supported the assignment of Glu138 as the general acid-base catalyst. Alanine replacements are prepared by site-directed mutagenesis to produce the inactive E138A and E134A mutants. Addition of azide reactivates the mutants in a concentration-dependent manner using an activated 2, 4-dinitrophenyl glycoside substrate. The chemical rescue operates by a different mechanism depending on the mutant as deduced from 1H NMR monitoring and kinetic analysis of enzyme reactivation. E138A yields the beta-glycosyl azide product arising from nucleophilic attack of azide on the glycosyl-enzyme intermediate, thus proving that Glu138 is the general acid-base residue. Azide activates the deglycosylation step (increasing kcat), but it also has a large effect on a previous step (as seen by the large decrease in KM, the increase in kcat/KM, and the pH dependence of activation), probably increasing the rate of glycosylation through Bronsted acid catalysis by enzyme-bound HN3. By contrast, azide reactivates the E134A mutant through a single inverting displacement to give the alpha-glycosyl azide product, consistent with Glu134 being the catalytic nucleophile. Formate as an exogenous nucleophile has no effect on the E138A mutant, whereas it is a better activator of E134A than azide. Although the reaction yields the normal hydrolysis product, a transient compound was detected by 1H NMR, tentatively assigned to the alpha-glycosyl formate adduct. This is the first case where a nonmodified sugar gives a long-lived covalent intermediate that mimics the proposed glycosyl-enzyme intermediate of retaining glycosidases.