Previous results have indicated the involvement of the hippocampus in the behavioral effect of vasopressin, with a better effect when the peptide was injected in the ventral part rather than in the dorsal part of this structure. The purpose of the present study was to determine, in mice, whether the injection of vasopressin or vasopressin antisera into the ventral hippocampus has an effect on retrieval and relearning of a Go-No Go visual discrimination task and, if so, to what extent this involvement of the vasopressin system depends on the integrity of the medial amygdaloid nucleus, the main source of vasopressin innervation in the ventral hippocampus in rats. In the first experiment, we showed that pretest microinjection of Arg8-vasopressin (25 pg per animal) in the ventral hippocampus alleviated forgetting observed after a prolonged interval of 24 days between the acquisition of information and its retrieval. This enhancing effect was characterized by better retrieval and relearning in vasopressin-treated mice than those in control mice. Conversely, an immunoneutralization of endogenous vasopressin in the ventral hippocampus by the microinjection of vasopressin antisera (1/10 dilution) resulted in the drastic impairment of retrieval and relearning. Since the lack of an observable change in a locomotor activity test might explain these results, we postulated that the vasopressin system in the ventral hippocampus is involved in retrieval processes. Moreover, the effects of these treatments in a nonassociative context suggest that the effect of vasopressin could be dependent on the contextual paradigm used. In the second experiment, we localized vasopressin immunoreactive fibers in the CA1-CA2 ventral hippocampal fields and CA4-gyrus dentatus region, and vasopressin perikarya in the medial amygdaloid nucleus. Then, the projection of vasopressin cells from the medial amygdaloid nucleus to the ventral hippocampus was evaluated by studying changes in vasopressin immunoreactive fiber density in the ventral hippocampus after a lesion of the medial amygdaloid nucleus. The results showed the almost complete disappearance of vasopressin fibers in the CA1-CA2 hippocampal fields after the medial amygdaloid lesion. In contrast, vasopressin fibers in the CA4 and gyrus dentatus region remain unchanged. On the basis of our immunohistochemical results, our third experiment tested the repercussions of the change in vasopressin innervation in the ventral hippocampus, due to the medial amygdaloid lesion, on the effects of exogenously administered vasopressin on both retrieval and relearning processes. The medial amygdaloid lesion induced a deleterious effect on retrieval without really affecting the ability to relearn. No observable change in locomotor activity could explain this impairment.(ABSTRACT TRUNCATED AT 400 WORDS)