Thermal conductivity data for crude (mainly heavy) oils and mixtures of crude oils and pure hydrocarbons were collected at temperatures from 20 to 125 degrees C and pressures up to 10 MPa using a hot wire apparatus. A criterion was established to screen out data that were affected by convection. The screened data and data from the literature were used to develop an Expanded Fluid (EF) based thermal conductivity model for pure hydrocarbons, crude oils and their mixtures. The proposed model is applicable across the entire phase diagram including the critical region; however, it does not predict the critical enhancement of thermal conductivity observed in the vicinity of the critical point. The model inputs are the density of the fluid, the pressure, the dilute gas thermal conductivity, the compressed state density, and three other fluid specific parameters. The gas thermal conductivity is calculated from a well-established correlation. The compressed state density is obtained from the literature or from fitting the EF viscosity model to viscosity data. The three fluid specific parameters are determined by fitting the model to thermal conductivity data. The model fits the data of 63 pure hydrocarbons at temperatures from -150 to 330 degrees C and pressures up to 200 MPa with average deviation of 4%, except in the vicinity of the critical point (0.97 < reduced temperature < 1.1). The model also fits the data of 7 different crude oils at temperatures and pressures up to 150 degrees C and 10 MPa to within 0.3% of the experimental data. Mass based mixing rules were proposed for the model parameters of mixtures. The data for 19 pure hydrocarbon binaries at atmospheric pressure were predicted with average deviation of 0.5% and that of 8 bitumen/solvent pseudobinaries, at pressures up to 10 MPa, was predicted with average deviation of 1.6%. The introduction of binary interaction parameters into the mixing rules halved the magnitude of the deviations.