In hydrothermal studies, the depth of the hydrothermal system is always required, but rarely known via traditional geophysical exploration techniques. While previous studies have shown that continuous wavelet transform algorithms applied to self-potential data can theoretically determine the depth of the hydrothermal fluids, this study uses multi-scale wavelet tomography with multiple wavelets, field measurements and geophysical models to accurately determine this depth. On Stromboli, Waita and Masaya volcanoes, multi-scale wavelet tomography of field measurements gives reproducible depth results, supported by independent geophysical measurements and models, and accurately locates the main water flow paths at shallow depths. Unlike other traditional geophysical methods, multi-scale wavelet tomography using self-potential data is a low cost tool to rapidly determine depths of the shallowest hydrothermal structures. This approach has the potential to significantly enhance our ability to locate geothermal systems and monitor active volcanoes.