In a series ofinvivoexperiments on five adult canines, a small cylindrical permanent magnet (∼5‐mm diameter×5 mm long) was magnetically moved under fluoroscopic guidance from an occipital‐lobe burr hole to a predetermined destination within the brain and then removed. On three of the animals, dorsal and temporal skull markers were used to establish a coordinate system against which the motions of the seed were referenced. These procedures were sufficiently accurate to permit the guided motion of the seed along nonlinear paths within the brain, including traversal of the midline through the corpus callosum. For removal, the seed could be steered either to a frontal lobe location for extraction through an auxiliary burr hole, or back to the same burr hole through which it had been inserted. This article discusses the way in which stereotactic motions were obtained, the performance limits of the instrumentation and the precision of motion achieved.