The $\beta$-glucosidase (EC.3.2.1.21), a member of cellulolytic enzyme system, was isolated from the culture filtrate of $\mbox{\underline{Aspergillus}}$ $\mbox{\underline{phoenicis}}$ QM 329, and was partially purified by DEAE-cellulose ion-exchange chromatography and by an affinity column employing p-aminophenyl-l-thio-$\beta$-D-glucoside as a ligand on Sepharose-4B matrix. The enzyme partially purified showed a very limited substrate specificity for $\beta$-D-glucoside. The Km value for the p-nitrophenyl-$\beta$-D-glucoside was 0.75 mM and it showed substrate inhibition above 5 mM with p-nitrophenyl-$\beta$-D-glucoside and cellobiose. The glycone moieties of substrates (e.g. $\beta$-D-glucose) were essential for substrate binding and the aglycone moiety had little specificty for enzyme-substrate binding. And the activity of $\beta$-glucosidase was optimal at PH 4.5 and 65$^\circ$C, and energy of activation for pnitrophenyl-$\beta$-D-glucoside was 11.2 Kcal/mole. Addition of polyols (especially, L-arabitol) to reaction mixture showed increment of pnitrophenol and decrement of $\beta$-D-glucose production with p-nitrophenyl-$\beta$-D-glucoside as a substrate. Their kinetic modes were expected from derived equations and transglucosylation product was detected with paper chromatography. Nojirimycin was a potent inhibitor, of which Ki value was 0.24 $\mu$M and hydroxylamine showed mixed type inhibition in acetate buffer or pyridine buffer. Heavy metals such as $Hg^{++}$ and $Ag^+$ and p-chloromecuribenzoate showed not marked inhibition, but bromosuccinimide showed strong inhibition. From kinetic analysis of PH-rate studies and carboxyl group modification study, there might be 2 carboxyl groups with $pk_1 = 4.1$ and $pk_2 = 6.1$. And a new model of $\beta$-glucosidase reaction mechanism was proposed.