Studies of brain-behaviour relationships in hippocampal amnesia largely ignore the presence and explanatory potential of knock-on effects beyond the medial temporal lobes. In a large cohort of patients (n=38) with hippocampal damage due to autoimmune limbic encephalitis, we had reported evidence that extra-hippocampal structural and functional abnormalities in the broader hippocampal-diencephalic-cingulate network fully mediated the effects of hippocampal damage on several aspects of episodic memory. However, we had not examined the specific diencephalic nuclei affected or the white matter pathways that would help explain these remote effects. In this study, we used recently developed methods of automated segmentation of diencephalic nuclei, as well as a range of analyses of white matter integrity. As expected, we found atrophy in the anterior thalamic nuclei and in the mammillary bodies, but also in the laterodorsal, pulvinar, and dorsomedial nuclei. The extent of atrophy in some of these nuclei was comparable to, if not larger than that observed for the hippocampal formation, even though none of our patients\' acute clinical scans disclosed thalamic damage. We also present evidence linking these volumes to patients\' episodic memory impairment over and above any correlation with hippocampal/subicular subfield volumes. White matter integrity was strongly compromised in the hippocampal-diencephalic-cingulate network, and the volumetric relationship between the hippocampal formation and the mammillary bodies was at least partly mediated by the integrity of the fornix across patients, consistent with the assumption of Wallerian degeneration following focal medial temporal lobe damage. However, evidence for the specialisation of the hippocampal-diencephalic-cingulate network in recollection/recall was mixed, and white matter abnormalities extended to regions well beyond this network. Our findings highlight the need to longitudinally examine diencephalic and white matter integrity in cohorts of hippocampal damage with different aetiologies and update neuroanatomical models of this network.