Since the description of Isisfordia duncani, a number of new extinct species and revisions of previous species have prompted a variety of contradicting phylogenetic hypotheses on the 1 topology of the Neosuchia. As a consequence, a consensus on the rooting of Eusuchia in relation to other neosuchian clades has not been reached and the origin of the group remains unsettled. Exemplifying this, Bernissartia fagesii from the Early Cretaceous of Belgium has long been considered a key taxon for understanding the origin of Eusuchia, but more recent hypotheses found support for a more basal position, as an ally to goniopholidids, paralligatorids or atoposaurids. Because many details of the anatomy of the type specimen are hidden by glue and sediment still adhering to the fossils, a number of characters are pending confirmation. Based on computed tomography data, we extract bones of the cranium and mandibles, describe new characters and re-evaluate anatomical details in the lectotype specimen. Our phylogenetic analysis confirms that B. fagesii is a derived neosuchian, unrelated to atoposaurids, goniopholidids and paralligatorids. We recover B. fagesii and Koumpiodontosuchus aprosdokiti in a basal position within Eusuchia, together with Susisuchidae, a group of gondwanan neosuchians containing Susisuchus and Isisfordia, which are here forming a polytomy with Hylaeochampsidae. The presence-absence of pterygoidbound internal choanae cannot be used to fully resolve relationships at the neosuchianeusuchian transition because of the variability of this character even at the familial level as recently reported within susisuchids and bernissartiids. There is no doubt that true eusuchians were present in Laurasia as early as the Early Cretaceous, the hylaeochampsid Hylaeochampsa vectiana being the oldest (Barremian) undoubted representative. But whether the Eusuchia were also present in southern landmasses depends on solving the phylogenetic position of susisuchids and other less known gondwanan forms within or outside Eusuchia.