Land use and land cover (LULC) exert strong controls on soil properties and thus water flow in soils. The question arises whether existing models designed for simulating water and solute movement in heterogeneous soil include the effects of LULC on infiltration and percolation. To answer this question, the predictive capacity of two solute transport models was benchmarked to simulate flow processes in soils developed under different LULCs. To benchmark the model performance, dye tracer sprinkling experiments were conducted on five sites displaying similar soil texture and parent material but differences in LULC (two farmland sites, tilled and untilled, two grassland sites, and one deciduous forest site). Water content changes were continuously measured at different depths with time domain reflectometers during 60-mm irrigation over 4 h. After the tracer application, vertical and horizontal soil sections were excavated and photographed. The experimental data were then compared with the simulations based on a multicriteria benchmarking strategy, including simulated vs. observed water content changes, solute distribution profiles, and maximum infiltration depth. Both models were capable of predicting water flow for one of the grassland sites but revealed severe weaknesses when additional flow processes not explicitly included in the model structure came into play. Poor model outcomes resulted from LULC effects such as a strong surface microtopography altering macropore flow initiation in the agricultural soils and from horizontally oriented roots and surface water repellency in the forest soil. Our results suggest that LULC effects need to be better incorporated into the conceptualization and parameterization of infiltration and percolation in hydrologic models to obtain realistic predictions concerning water quality and quantity.