A tomographic metal detection and characterization system has been designed and built for recovering information about magnetic and/or conductive objects within the detector space. This information is gathered as a result of a ‘walk-through’ scan of a candidate in the same manner as for a typical security metal detector archway. Following the passage of the candidate, the system uses measurements from an array of coils to calculate the polarizability tensor, which describes the low frequency electromagnetic characteristic of a small metallic object when it interacts with an ac magnetic field. In addition to the magnetic polarizability dyadic tensor, the position of the perturbation is also determined as a product of the inversion algorithm. The system has been tested and is capable of inverting object tensors with <20% typical parameter variation, and determines three-dimensional object location with a typical error of less than ±3 cm. In this paper, results are shown from a set of four different test object examples, each with a different magnetic polarizability tensor. This object set consists of a ferrite sphere, a ferrite rod and phantom aluminium and steel handgun shapes.