Homing in the box turtle, Terrapene Carolina, has been well documented, but the mechanisms involved are largely unknown. We used laboratory and field experiments to determine if the geomagnetic field may be used as an orientational cue. Turtles were trained in the laboratory to move along an east/west axis and were tested for directionality in a circular arena. Control turtles with dummy weights attached to the carapace oriented significantly to the training direction. Experimental turtles with magnets attached to the carapace were randomly oriented. In field tests, turtles were displaced from their home ranges to 4 release sites located symmetrically about the home site. Controls were displaced under normal magnetic field strength (Helmholtz coils off) and experimentals were displaced under increased field strength (15% increase over local geomagnetic field strength). The initial orientation of the control turtles clustered significantly in the homeward direction while the initial orientation of experimental turtles was not clustered in the homeward direction. Both control and experimental turtles showed compass orientation clustered in an eastward direction regardless of the directional relationship between the release site and home. This compass orientation appears to mask, to some degree, the homeward orientation of the turtles.