Involvement of the medial septal area (MSA) in reference memory and working memory versions of the Morris water maze (MWM) task was investigated in rats with reversible inactivation of this area by drugs injected through a single cannula aimed at the MSA. In Experiment 1, rats were trained in a reference memory version of the MWM with two blocks of four trials per day for 3 consecutive days. Acquisition was impaired by pretrial MSA injection of 10 ng tetrodotoxin (TTX) in 1 microliter saline but not of saline alone into MSA. In Experiment 2, intraseptal injection of TTX (10 ng, 1 microliter) immediately after two blocks of four trials had no effect on the consolidation of spatial reference memory. In Experiment 3, intraseptal injection of TTX (10 ng, 1 microliter) impaired retrieval of well established spatial reference memory in rats which had received 8 trials per day for 3 consecutive days. In Experiments 4 and 5, rats were trained in a working memory version of MWM task to find a new target position in trial 1 and retrieval of this information was tested 75 min later in trial 2. Intraseptal injection of lidocaine (4%, 1 microliter) prior to training impaired working memory performance while immediately posttraining injection of lidocaine had no effect. It is concluded that normal activity of the MSA is necessary for the memory formation at the time of training but its involvement in posttraining consolidation is unlikely. The MSA function is required for retrieval of well established spatial reference memory.