New neurons are generated daily in the hippocampus during adult life. They are integrated into the existing neuronal circuits according to several factors such as age, physical exercise and hormonal status. At present, the role of these new neurons is debated. Computational simulations of hippocampal function allow the effects of neurogenesis to be explored, at least from a computational perspective. The present work implements a model of neurogenesis in the hippocampus with artificial neural networks, based on a standard theoretical model of biologically plausible hippocampal circuits. The performance of the model in retrieval of a variable number of patterns or memories was evaluated (episodic memory evaluation). The model increased, in a phase subsequent to initial learning, the number of granular cells by 30% relative to their initial number. In contrast to a model without neurogenesis, the retrieval of recent memories was very significantly improved, although remotes memories were only slightly affected by neurogenesis. This increase in the quality of retrieval of new memories represents a clear advantage that we attribute to the neurogenesis process. This advantage becomes more significant for higher storage loads. The model presented here suggests an important functional role of neurogenesis on learning and memory. (Contains 7 figures.)