Résumé : An ERP study on 9 healthy participants was carried out to temporally constrain the neural network proposed by Campanella et al. (2001) in a PET study investigating the cerebral areas involved in the retrieval of face-name associations. Three learning sessions served to familiarize the participants with 24 face-name associations grouped in 12 male/female couples. During EEG recording, participants were confronted with four experimental conditions, requiring the retrieval of previously learned couples on the basis of the presentation of name-name (NN), face-face (FF), name-face (NF), or face-name (FN) pairs of stimuli. The main analysis of this experiment consisted in the subtraction of the nonmixed conditions (NN and FF) from the mixed conditions (NF and FN). It revealed two main ERP components: a negative wave peaking at left parieto-occipital sites around 285 ms and its positive counterpart recorded at left centro-frontal electrodes around 300 ms. Moreover, a dipole modeling using three dipoles whose localization corresponded to the three cerebral areas observed in the PET study (left inferior frontal gyrus, left medial frontal gyrus, left inferior parietal lobe) explained more than 90% of the variance of the results. The complementarity between anatomical and neurophysiological techniques allowed us to discuss the temporal course of these cerebral activities and to propose an interactive and original anatomo-temporal model of the retrieval of face-name associations.