Low-cost saccharification is one of the key bottlenecks hampering the further application of lignocellulosic biomass. Clostridium thermocellum is a naturally ideal cellulose degrading bacterium armed with cellulosomes, which are multienzyme complexes that are capable of efficiently degrading cellulose. However, under controlled condition, the inhibition effect of hydrolysate cellobiose severely restricts the hydrolytic ability of cellulosomes. Although the addition of beta-glucosidase (Bgl) could effectively relieve this inhibition, the spatial proximity effect of Bgl and cellulosomes on cellulose degradation is still unclear. To address this issue, free Bgl from Caldicellulosiruptor sp. F32 (CaBglA), carbohydrate-binding module (CBM) fused CaBglA (CaBglA-CBM) and cellulosomal type II cohesin module (CohII) fused to CaBglA (CaBglA-CohII) were successfully constructed, and their enzymatic activities, binding abilities and saccharification efficiencies were systematically investigated in vitro and in vivo. In vivo, with the adjacency of CaBglA to cellulosomes, the saccharification efficiency of microcrystalline cellulose increased from 40% to 50%. For the pretreated wheat straw, the degradation rate of the combination of cells and the CaBglA-CohII or the CaBglA-CBM was as efficient as that of the free CaBglA (approximately 60%). This study demonstrated that the proximity of CaBglA to cellulosomes had a positive effect on microcrystalline cellulose but not on pretreated wheat straw, which may result from the nonproductive adsorption of lignin and the decreased thermostability of CaBglA-CBM and CaBglA-Cohn compared to that of CaBglA. The above results will contribute to the design of cost-effective Bgls for industrial cellulose degradation.