This paper presents an evaluation of various heavy oil viscosity models using new experimental data for mixtures of a heavy oil sample from Canada's heavy oil reserves and n-decane used as a viscosity reducer. We measured viscosities at temperatures ranging from ambient to 450 K using a versatile mercury capillary tube.We evaluated several mixing rules to predict the mixture viscosity as a function of temperature and the viscosities of the constituents. The most promising mixing rule follows a methodology proposed originally by Lederer. We used our own experimental data to calibrate this mixing rule, which only has two adjustable parameters, and used the same coefficients to predict the blending viscosity of mixtures of bitumen and a synthetic oil solvent reported in the literature. This mixing rule can reproduce blending viscosities within the uncertainty of the experiments for oil-solvent viscosity ratios up to 4 x 10(5), temperatures from ambient to 450 K, and mixtures of any solvent proportion.This paper provides a valuable new experimental viscosity of heavy oil and n-decane and a reliable mixing rule that has been independently validated with literature data and can be used to estimate diluent quantities required to reduce oil viscosity for pipeline transportation of heavy oil, to design solvent stimulation of wells, and to provide viscosities for thermal reservoir simulators for heavy oil recovery processes. (C) 2003 Elsevier B.V. All rights reserved.