Navigation by means of cognitive maps appears to require the hippocampus; hippocampal place cells (PCs) appear to store spatial memories because their discharge is con¢ned to cell-speci¢c places called ¢ring ¢elds (FFs). Experiments with rats manipulated idiothetic and landmark-related information to understand the relationship between PC activity and spatial cognition. Rotating a circular arena in the light caused a discrepancy between these cues. This discrepancy caused most FFs to disappear in both the arena and room reference frames. However, FFs persisted in the rotating arena frame when the discrepancy was reduced by darkness or by a card in the arena. The discrepancy was increased by`¢eld clamping' the rat in a room-de¢ned FF location by rotations that countered its locomotion. Most FFs dissipated and reappeared an hour or more after the clamp. Place-avoidance experiments showed that navigation uses independent idiothetic and exteroceptive memories. Rats learned to avoid the unmarked footshock region within a circular arena. When acquired on the stable arena in the light, the location of the punishment was learned by using both room and idiothetic cues; extinction in the dark transferred to the following session in the light. If, however, extinction occurred during rotation, only the arena-frame avoidance was extinguished in darkness; the room-de¢ned location was avoided when the lights were turned back on. Idiothetic memory of room-de¢ned avoidance was not formed during rotation in light; regardless of rotation, there was no avoidance when the lights were turned o¡, but room-frame avoidance reappeared when the lights were turned back on. The place-preference task rewarded visits to an allocentric target location with a randomly dispersed pellet. The resulting behaviour alternated between random pellet searching and target-directed navigation, making it possible to examine PC correlates of these two classes of spatial behaviour. The independence of idiothetic and exteroceptive spatial memories and the disruption of PC ¢ring during rotation suggest that PCs may not be necessary for spatial cognition; this idea can be tested by recordings during the place-avoidance and preference tasks.