In 1941, Abel established Balaena belgica based on a series of fused cervical vertebrae and citing other cranial fragments from the late Neogene of the Antwerp harbor (northern Belgium). Later, Plisnier-Ladame & Quinet (1969) added a neurocranium and other skeletal remains from the same area to this species. Recently, the neurocranium was re-assigned to the genus Eubalaena thanks to newer phylogenetic analyses. Here, a new description is provided of materials previously assigned to “Balaena” belgica together with taxonomic revisions. Our work suggests that the cervical complex originally designated as the type of “Balaena” belgica is too poorly preserved to be used as such and is assigned to Balaenidae gen. et sp. indet., thus making “Balaena” belgica a nomen dubium. In addition to the neurocranium, the other remains consist in a fragment of maxilla assigned to Balaenidae gen. et sp. indet. and in a humerus assigned to Eubalaena sp. Discovered in the Kruisschans Sands Member of the Lillo Formation (3.2–2.8 Ma, Piacenzian, Late Pliocene), the neurocranium is designated as the holotype of the new species Eubalaena ianitrix. Our phylogenetic analysis supports a sister-group relationship of Eubalaena ianitrix and Eubalaena glacialis, and helps constraining the ages of origin for balaenid clades. Ecological and phylogenetic data suggest that Eubalaena ianitrix may represent the direct ancestor of Eubalaena glacialis, the latter having evolved through phyletic transformation including body size increase during the temperature decline of the Late Pliocene.