In this study, the functional properties of the medial prefrontal cortex (mPFC) of the rat were examined in a task for spatial navigation. The dorsomedial (dmPFC) and ventromedial (vmPFC) parts of the PFC differ in their connectivity with the hippocampus, vmPFC being the main recipient of hippocampal input. Therefore, it is hypothesized that in a spatial task known to be particularly sensitive to hippocampal, but also to prefrontal cortical damage, especially vmPFC is committed to proper spatial learning and memory. Using the Morris water maze task, with an invisible platform, male rats with either partial (dmPFC or vmPFC) or whole (mPFC) bilateral lesions were subjected to various procedures reflecting spatial learning and memory, including a spatial reversal. Animals with dmPFC, vmPFC and mPFC lesions learned and remembered the spatial task equally well as their controls, regardless of the size or site of the bilateral lesion. However, when presented with a spatial reversal, animals with whole mPFC lesions were initially slower in locating the invisible platform than controls and animals with partial mPFC damage, but this effect was only short-lasting. When subsequently presented with a visible platform, all animals with partial (dmPFC or vmPFC) or whole mPFC damage needed significantly more time to locate the platform than the controls. This effect was transient and lasted longer in the animals with whole mPFC damage than in animals with partial damage. On the basis of these findings we conclude that an intact mPFC is not necessary for proper spatial learning and memory. The impairments in the reversal task and especially in the visually-cued task can best be explained as a diminished behavioural flexibility when a shift in task demands occurs. Since the degree of this impairment was related to the size of these lesions, but not to their sites, these differences are ascribed to a 'mass action' of medial prefrontal cortex lesions.