Viscosity is an important transport property for engineering design and simulation of bitumen production and transportation. During the production of bitumen with solvent injection, steam-assisted gravity drainage, or expanding solvent steam-assisted gravity drainage, the oil/solvent mixture encounters various temperature and pressure conditions. Therefore, a model is necessary to predict the viscosity of the mixture of bitumen and solvent in wide ranges of temperatures, pressures, and compositions (T-P-x). In this work, we propose a semitheoretical viscosity model based on the Arrhenius mixing rule and considering the effect of association between the molecules of the solvent and the bitumen. To achieve this purpose, thermodynamic perturbation theory (TPT) is used to calculate the fraction of bonding solvent molecules. We calculate the viscosity of the solvent in wide temperature and pressure ranges using the modified Enskog theory (MET). Results show an acceptable agreement between the predictions of this model and experimental viscosity data of bitumen saturated with different solvents (CH4, N2, CO2, and C2H6) at various T-P-x ranges. These experimental data cover the typical T-P-x ranges of oil-recovery methods.