Surgical placement of implantable medical devices (IMDs) has limited precision and post-implantation the device can move over time. Accurate knowledge of the position of IMDs allows better interpretation of data gathered by the devices and may allow wireless power to be focused on the IMD thereby increasing power transfer efficiency. Existing positioning methods require device sizes and/or power consumptions which exceed the limits of in-vivo mm-sized IMDs applications. This paper describes a novel implant positioning system which replaces the external transmitting (TX) coil of a wireless power transfer link by an array of smaller coils, measures the mutual inductance between each coil in the TX array and the implanted receiving (RX) coil, and uses the spatial variation in those mutual inductances to estimate the location of the implanted device. This method does not increase the hardware or power consumption in the IMD. Mathematical analysis and electromagnetic simulations are presented which explain the theory underlying this scheme and show its feasibility. A particle swarm based algorithm is used to estimate the position of the RX coil from the measured mutual inductance values. MATLAB simulations show the positioning estimation accuracy on the order of 1 mm.