Infection caused by hepatitis C virus (HCV) is a significant world health problem for which novel therapies are in urgent demand. Nonstructural (NS5B) viral proteins have emerged as an attractive target for drug discovery efforts toward antiviral for hepatitis C virus. Toward this target several series of NS5B inhibitors that showed activity in the replicon assay have been reported. In this article, we gave a report of the NS5B allosteric sites and the corresponding non-nucleoside inhibitors, which belong to different chemical classes. Then using comparative molecular field analysis (CoMFA) and comparative molecular similarity indices analysis (CoMSIA) methods, 3-dimension quantitative structure-activity relationships (3D-QSAR) models have been built with more than two hundred benzimidazole/indole derivative inhibitors. These studies indicated that the QSAR models were statistically significant and had high predictabilities (CoMFA: q(2)=0.823, r(2)=0.942; CoMSIA: q(2)=0.817, r(2)=0.935). The flexible docking method, which was performed by the DOCK6.0 software, positioned all of the inhibitors into the allosteric site to determine the probable binding conformation. The CoMFA and CoMSIA models based on the docking conformations also yielded statistically significant and high predictive QSAR models (CoMFA: q(2)=0.509, r(2)=0.768; CoMSIA: q(2)=0.582, r(2)=0.854). Our models would offer help to better comprehend the structure-activity relationships existent for this class of compounds and also facilitate the design of new inhibitors with good chemical diversity.