We used time-lapse electrical resistivity tomography (ERT) and time-domain reflectometry (TDR) probes to noninvasively capture three-dimensional solute transport during four tracer experiments under different steady-state irrigation rates in a large unsaturated undisturbed soil column (140 cm length and 116 cm inner diameter). The transport was characterized by means of apparent convection–dispersion parameters that were derived from breakthrough curves (BTCs) at different lateral scales: the ERT voxel scale, the sampling volume of TDR, and the cross-section of the column. We validated the ERT-derived data by means of mass balance, TDR probes, and the effluent BTC. We observed an excellent mass recovery by ERT. The ERT-derived transport velocities exhibited minimal bias and high precision at the scale of the TDR measurements. On average the ERT-derived column-scale transport velocities were also not biased; however, the spatial variability of the voxel-scale velocities within the column's cross-sections underestimated the true velocity variability. In contrast to the transport velocities, the ERT-derived dispersivities exhibited a large bias and low precision and were sensitive to temporal smoothing. Unlike previous studies, we did not find evidence that the ERT-derived voxel-scale dispersivities increase with decreasing ERT sensitivity. Although ERT provided unprecedented information about transport processes, resolution and uncertainty analyses remain important issues requiring further investigation.