Upward water flow induced by evaporation can cause soil salinization and transport of contaminants to the soil surface and influences the migration of solutes to the groundwater. In this study, we used electrical resistivity tomography (ERT) to obtain time-lapse images of an upward-flow tracer experiment under evaporation conditions in a three-dimensional, spatially correlated heterogeneous laboratory soil composed of three different materials (coarse-, medium-, and fine-grained sands). The tracer experiment was performed during 40 d of quasi-steady-state, upward-flow conditions. Monitored transport was compared with three-dimensional numerical simulation based on the Richards and advection–dispersion equations. The ERT-derived and modeled solute transport correlated well in the lower part of the laboratory soil, while deviations increased toward the surface. Inversion of synthetic ERT data indicated that deviations cannot be explained by ERT data and inversion errors only, but also errors of the flow and transport model must be invoked. The classical potential/actual evaporation (Epot/Ea) concept underestimated the experimental evaporation, as locally Ea exceeded Epot, which was determined as the maximum evaporation from an insulated free water table minus soil heat flux. Increasing the potential evaporation rate uniformly in the model, so that wet high-evaporation zones can compensate for lower evaporation from dry zones, increased the correlation between experiment and model. Despite the remaining deviations, experiment and model showed a consistent and systematic pattern of preferential upward transport pathways. Close above the water table, most of the transport occurred in the coarse material, while with increasing height, transport was dominated by the finer materials. This study is an experimental benchmark for three-dimensional flow and transport models using simplified evaporation boundary conditions and for ERT to monitor upward transport.