The fate and transport of pollutants in the subsurface is usually predicted using numerical simulations. Comparisons of the simulation results with experimental data are required to validate the numerical codes. In this paper, the nonaqueous phase liquid (NAPL) simulator code was used to numerically simulate two centrifugal light nonaqueous phase liquid (LNAPL) transport experiments. The numerical predictions were compared with the centrifugal test results and the impact of capillary hysteresis was evaluated. It was concluded that the simulation with the nonhysteretic model slightly underestimated the LNAPL volume retained in the vadose zone, and the average error of the predicted LNAPL saturation in the vadose zone and along the plume centerline was approximately 5% to 10%. The application of capillary hysteresis for the numerical simulation of the second centrifuge test has eliminated this error. In addition, accounting for hysteresis in the numerical simulation of the second centrifuge test has considerably improved the predicted results with respect to the size of the lens-shaped plume within the capillary fringe. However, there were inconsistencies between the numerical and experimental results, especially with respect to the time for the LNAPL to reach the groundwater level. This was found to be due to overestimation of the simulated downward infiltration of the LNAPL in the capillary fringe. It was found that the pore-connectivity parameter with a value of 0.66 for the relative permeability–saturation model produced the best comparison with experimental results.