Modern irrigation techniques require accurate, rapid, cost-effective, spatial measurement of soil moisture. Electrical Resistivity Tomography (ERT) meets most of these requirements, but needs to be calibrated for each use because it is very sensitive to differences in soil characteristics. In this study, a pedotransfer function approach is used to remove the need for site-specific calibration, allowing ERT to be used directly to measure soil moisture. The study site was a hillslope vineyard, where eight calcaric-cambisol soil profiles were identified. From 2012 to 2013, 23 000 soil volume wetness measurements were acquired by Time Domain Reflectometry, and over 100 000 electrical resistivity data were collected in 160 ERT acquisitions. To better understand the ERT signal, soil texture, gravel content, cation exchange capacity, CaCO3, pH, organic carbon and total nitrogen were analysed in 64 soil samples from the study site. The sensitivity of ERT to differences in soil characteristics makes it difficult to establish a unique model linking electrical resistivity and soil moisture in heterogeneous soils. This study presents two possible solutions to overcome this problem, which are differentiated by the availability of data on soil characteristics. When these data are not available, it is possible to fit a number of different models for each homogeneous soil layer, but a site-specific calibration is necessary at least once. Conversely, when soil characteristics are available, they can be integrated into the model to build a pedotransfer function. A unique, accurate model is obtained in this way for all samples. In soils with similar characteristics to those observed, the function can be used directly to measure soil moisture by ERT. Developing pedotransfer functions such the one presented here could greatly improve, simplify and develop the use of electrical resistivity to measure soil moisture.