Finding and utilizing reliable water supplies for desert communities is important but difficult and expensive. This is especially the case in the remote north and northwest of South Australia, where there are a number of small Aboriginal communities with populations of less than a hundred people. Rainfall is sparse and intermittent (less than 200 mm/yr), and residents rely on groundwater for nonpotable supplies. Previous drilling in shallow sediments aiming for porous aquifers has led to problems with long-term well production. More recent wells targeting deeper fractured rock aquifers are showing more promise; unfortunately, such boreholes are more expensive and difficult to install. The aim of our work is to use natural-source magnetotelluric (MT) imaging to try and identify better sites for bores, and therefore reduce the risk of drilling a dry well. We hope to do this by imaging the 3D conductivity structure, allowing the identification of possible aquifers, and also by measuring the anisotropy due to water-bearing fractures, through the sensitivity of natural-source MT responses to electrically anisotropic layers in the subsurface. The first phase of our work, described here, is to test the method at a site where the hydrogeology is relatively well understood. The second phase will apply it to a community in need.