A broad-specificity beta-D-glucosidase from pig kidney cortex was isolated and purified to homogeneity by a rapid purification procedure. The pI (5.14 +/- 0.05), Mr (59,000 +/- 2000) and specific activities with several p-nitrophenyl glycosides (galactopyranoside, glucopyranoside, arabinopyranoside, xylopyranoside) were comparable with those published previously for cytoplasmic beta-D-glucosidase from other sources and organs. Mixed-substrate experiments and inhibition studies with glucono-(1----5)-lactone revealed that a single active centre, containing one catalytic site and one saccharide-binding site, was responsible for the splitting of all four synthetic substrates. Inhibition experiments with substrate analogues demonstrated that (i) the major binding determinant of the glycosides was the aglycone moiety, (ii) an anionic side chain of the enzyme (probably a carboxy group) interacted with the glycosidic linkages and (iii) the properties of the aglycone significantly influenced the binding of the carbohydrate moiety. The inhibition constants of the p-nitrothiophenyl derivatives were in good agreement with the Km values of the corresponding substrates. Therefore the Michaelis constants could be regarded as true equilibrium constants (Ks). The 'three-point-attachment model' of the substrate splitting, proposed by Daniels [(1983) Ph.D. Dissertation, University of Pittsburgh] for the analogous liver enzyme, was applicable for beta-D-glucosidase from pig kidney too. The possible nature of the 'attachments' is discussed.