High viscosity is a major concern in the recovery of heavy oil and bitumen. Viscosity reduction could be achieved by mixing bitumen with solvents. Cragoe(1) and Shu(2) have developed widely used methods for liquid mixture viscosity predictions. However, in these two models, the viscosities or densities of the heavy oil/bitumen and solvents have to be known at some reference condition. Low field nuclear magnetic resonance (NMR) relaxometry is an effective, non-destructive alternative for determining the petrophysical properties of oil reservoirs. It has also been shown to successfully predict the viscosity of conventional oils, heavy oils, and mixtures of oils with solvents. In this paper, a regression model of experimental data, Cragoe(l), Shu(1), and NMR models are compared with experimental data which were obtained from four heavy oil/bitumen samples mixed with six solvents in different ratios. NMR-based predictions are found to be similar to those of the Shu(2) model and superior to the predictions of the Cragoe(l) model.