Magnetoreception has been demonstrated in all five vertebrate classes. In rodents, nest building experiments have shown the use of magnetic cues by two families of molerats, Siberian hamsters and C57BL/6 mice. However, assays widely used to study rodent spatial cognition (e.g. water maze, radial arm maze) have failed to provide evidence for the use of magnetic cues. Here we show that C57BL/6 mice can learn the magnetic direction of a submerged platform in a 4-armed (plus) water maze. Naïve mice were given two brief training trials. In each trial, a mouse was confined to one arm of the maze with the submerged platform at the outer end in a predetermined alignment relative to magnetic north. Between trials, the training arm and magnetic field were rotated by 180(°) so that the mouse had to swim in the same magnetic direction to reach the submerged platform. The directional preference of each mouse was tested once in one of four magnetic field alignments by releasing it at the center of the maze with access to all four arms. Equal numbers of responses were obtained from mice tested in the four symmetrical magnetic field alignments. Findings show that two training trials are sufficient for mice to learn the magnetic direction of the submerged platform in a plus water maze. The success of these experiments may be explained by: (1) absence of alternative directional cues (2), rotation of magnetic field alignment, and (3) electromagnetic shielding to minimize radio frequency interference that has been shown to interfere with magnetic compass orientation of birds. These findings confirm that mice have a well-developed magnetic compass, and give further impetus to the question of whether epigeic rodents (e.g., mice and rats) have a photoreceptor-based magnetic compass similar to that found in amphibians and migratory birds.