A comparison of ten two-phase relative permeability models is conducted using experimental, semi-analytical and numerical approaches. Relative permeabilities from the models are compared with data obtained from 12 steady-state experimental runs on Berea and Brown sandstones using different combinations of three white mineral oils and 2% CaCl2 brine. Relative permeability data from four of the experimental runs are used to predict the displacement performance under Buckley-Leverett conditions and the results are compared against those predicted by the models. Finally, waterflooding performances predicted by the models are analyzed at three different viscosity ratios using a twodimensional, two-phase numerical reservoir simulator. The model results are compared against the experimental data using three different criteria. The models predict relative permeability to oil, relative permeability to water and relative permeability ratio values with varying degrees of success. The Wyllie-Gardner, Honarpour and Fatt-Dykstra models predict the best values for the relative permeability ratios. The Wyllie-Gardner, Honarpour and Burdine models give better results than the other models under Buckley-Leverett type displacement conditions. The simulator runs show that all models are able to run without problems (for the given conditions) although the Naar-Henderson model yields significantly different results than the other models. The simulator data show that as the oil-water viscosity ratio increases, the difference of performance predicted by different models diminishes. Overall, the Wyllie-Gardner and Honarpour models are found to yield consistently better results than the other models at the experimental conditions.