In 1968, Milner (Neuropsychologia 6 (1968) 191) demonstrated a face-memory impairment in patients with right, but not left, temporal-lobe excisions. Because all the removals included lateral and inferior temporal neocortex together with amygdala, parahippocampal gyrus and varying amounts of hippocampus, a combined-lesion effect could not be ruled out. We therefore examined the contribution of right temporal structures to recognition of previously unfamiliar faces by repeating Milner's original study, testing patients who had undergone selective amygdalo-hippocampectomy (AH), in addition to those with anterior temporal-lobectomy (TL). The paradigm involved selecting 12 previously studied faces from an array of 25 photographs. The Mooney Closure Faces Test was also administered. Subjects included 29 AH patients (14 left (LAH) and 15 right (RAH)) and 59 TL patients (30 L and 29 R) who were categorized further based on extensive (18 LTH and 21 RTH) or minimal (12 LTh and 8 RTh) hippocampal encroachment. Twenty age- and education-matched normal control subjects (NC) were also tested. For the face-memory task, one-way ANOVA revealed a strong group effect (P<0.001), and post-hoc tests confirmed that both the RTH and RAH groups recognized fewer faces than the NC and LAH groups; the RAH group also differed from the LTh, LTH and RTh groups. No group differences were found for the closure test. Our findings suggest that right medial temporal-lobe structures are critically involved in the retention, but probably not in the perception, of new faces.