The network of prior streams and palaeochannels common across the Riverine Plains of the Murray–Darling Basin act as conduits for the redistribution of water and soluble salts beneath the root-zone. To improve scientific understanding of these hydrological processes there is the need to better represent and map the connectivity and spatial extent of these physiographic and stratigraphic features. Groundbased electromagnetic (EM) instruments, which measure bulk soil electrical conductivity (σ_a), have been used widely to map their areal distribution across the landscape. However, methods to resolve their location with depth have rarely been attempted. In this paper we employ a 1-D inversion algorithm with 2-D smoothness constraints to predict the true electrical conductivity (σ) at discrete depth increments using EM data. The EM data we use include the root-zone measuring EM38 and the deeper sensing EM34. We collected EM38 data in the vertical (EM38v) and horizontal (EM38h) dipole modes and EM34 data in the horizontal mode and coil spacing of 10, 20, and 40 m (respectively, EM34-10, EM34-20, and EM34-40). In order to compare and contrast the value of the various EM data we carried out multiple inversions using different combinations, which include: independent inversions of (i) EM38 (root-zone) and (ii) EM34 data (vadose-zone), and in combination using (iii) EM38v, EM38h, and EM34-10 (near-surface), and (iv) all 5 EM datasets (regolith) available. The general patterns of σ are shown to compare favourably with the known pedoderms, physiographic, and stratigraphic features and soil particle size fractions collected from calibration cores drilled across the lower Macquarie Valley study area. In general we find that the EM38 assists in resolving root-zone variability, specifically duplex soil profiles and physiographic features such as prior streams, while the use of the EM34 assists in resolving the stratigraphic nature of the vadose-zone and specifically the likely location of palaeochannels and subsurface anomalies that may indicate the location of good quality groundwater and/or clay aquitards. In this case, our potential to use σ to predict clay content is limited by the non-linearity of the cumulative functions. In order to improve on the non-linearity of our inversion we need to develop a full solution of the forward problem.