Numerical simulations of solute transport under unsaturated flow conditions were employed in order to test the capability of the upscaling methodology that Dagan developed for steady state, saturated flow and that Russo adapted to steady state, unsaturated flow to compensate for the loss of subgrid variations of the velocity field caused by coarse discretization of the flow domain. The results suggest that under relatively simple, steady state, gravity-dominated flows and for soils of differing textures the basic requirement for the upscaling of the dispersion coefficients is fulfilled and that the upscaling methodology essentially compensated for the loss of subgrid variations of the velocity field. For both soils these desirable results were achieved with a relatively coarse grid, which in turn, reduced the number of numerical cells by 96% compared with the fine-grid discretization of the flow domain. The applicability of the upscaling methodology to more general situations involving complex, transient flow regimes originating from periodic rain/irrigation events and water uptake by plant roots was also analyzed. The results of these analyses suggested that under transient flow conditions, in the absence of water uptake by plant roots, the basic requirement for the upscaling of the dispersion coefficients was fulfilled within numerical errors and the upscaling methodology essentially compensated for the loss of subgrid variations of the velocity field. On the other hand, in the case of water uptake by plant roots the failure of the numerical solution of the flow equation over coarse-grid cells to reproduce the actual complex flow pattern in the root zone prevented the fulfillment of three out of the four equalities which comprise the basic requirement for the upscaling of the dispersion coefficients. Nevertheless, even in this complicated case, the upscaling methodology essentially compensated for the loss of subgrid-scale variations of the velocity field caused by coarse discretization of the flow domain.