Physically based models are increasingly applied to analyze contaminant transport in soil by preferential water flow. Unfortunately, in the past, preferential flow models were rarely evaluated using appropriate experimental data because of the complexity of conceptual models and limitations of measuring techniques. In this study, we designed a novel soil column experiment with advanced measurement techniques that enabled us to discriminate macropore and matrix water flow and quantify interdomain (macropore-matrix) water transfer. Experiments of drainage and upward and downward infiltration revealed hydraulic nonequilibrium between matrix and macropore domains. Cumulative interdomain water transfer could be estimated using mass balance calculations. In a hierarchical modeling approach, four numerical models of different complexity were compared to the column experiment data. As a reference model, pseudo three-dimensional axisymmetric Richards' equation (ARE) was used for inverse estimation of domain-specific hydraulic parameters. The parameters were subsequently used for performance evaluation of an equivalent continuum model with bimodal hydraulic functions (ECM) and two dual-permeability models with first-order (DPM1) and second-order (DPM2) terms for water transfer between macropore and matrix. Overall, DPM2 gave slightly more accurate domain-specific results of water flow, water contents, and pressure heads than DPM1. Although bulk soil water flow results for the ECM were least accurate, they were considered to be within acceptable range. Compared to the more comprehensive ARE approach, DPMs were found to be almost equally capable of simulating interdomain and intradomain water flow and could be considered more versatile as they allow for a variety of macropore-matrix geometries.