The determination of thermal conductivity of reservoir materials is an important requirement for the proper evaluation of heat transfer in oil sand reservoirs. However, a correct assessment of such a property can result in the demand for a wide range of tests of various oil sand samples, often varying in grain sizes, statistics, porosities and saturations. This complexity of such factors, and the inadequate supply of accurate empirical data thus makes it difficult to obtain robust and correct mixing rules that can be used in numerical simulators. In the present study, the aim then is to conduct a systematic study into the effects of grain statistics, porosity and saturation on thermal conductivity, and to provide predictive models for such parameters. These aims are accomplished by conducting a series of thermal conductivity tests on oil sand packs whose particle size distributions, porosity and saturation are known. In the substantive thermal conductivity tests, samples of varying grain statistics are subjected to an axial heat transfer in a thermal conductivity apparatus. The heat power generated into the system is recorded together with an array of temperatures at specific locations within the apparatus. These data are then matched in a numerically simulated model of the apparatus in order to obtain the thermal conductivities of the samples. The thermal conductivities are then analyzed to determine the effects of grain statistics, porosity and saturation. The results indicate that the thermal conductivities are predominantly affected by water saturation. Other relationships with respect to core statistics and porosities are identified. The data supplied will serve an additional utility of validating numerical models.