Reversible, lidocaine-induced lesions of the CA1/subicular subfield of the ventral hippocampus or the shell region of the nucleus accumbens (N.Acc.) were used to assess the roles of these structure during the acquisition and retention of a spatial response as measured by the Morris water-maze task. Acquisition and retention tests were administered over 2 phases of 6 trials, respectively. Rats receiving reversible lesions of the ventral CA1/subiculum prior to the acquisition phase of this task required significantly longer path lengths to find a hidden platform than animals which received control infusions of artificial cerebrospinal fluid. Rats with similar lesions to the N.Acc. were unimpaired. During the retention phase, 30 min after the acquisition phase, rats with prior ventral CA1/subiculum or N.Acc. lesions had similar path lengths to control animals. Lidocaine infusions into either the ventral CA1/subiculum or N.Acc. prior to the retention phase did not impair performance relative to control animals. These results suggest that the N.Acc. is not involved in either the acquisition or retention of spatial information. In contrast, the ventral CA1/subiculum does appear to be involved in the initial use of novel spatial information necessary for the performance of a spatially mediated escape response, but is not involved in the retention or retrieval of previously acquired spatial information.