Brachiosauridae is a clade of titanosauriform sauropod dinosaurs that includes the well-known Late Jurassic taxa Brachiosaurus and Giraffatitan . However, there is disagreement over the brachiosaurid affinities of most other taxa, and little consensus regarding the clade’s composition or inter-relationships. An unnamed partial sauropod skeleton was collected from middle–late Oxfordian (early Late Jurassic) deposits in Damparis, in the Jura department of eastern France, in 1934. Since its brief description in 1943, this specimen has been informally known in the literature as the ‘Damparis sauropod’ and ‘French Bothriospondylus ’, and has been considered a brachiosaurid by most authors. If correctly identified, this would make the specimen the earliest known titanosauriform. Coupled with its relatively complete nature and the rarity of Oxfordian sauropod remains in general, this is an important specimen for understanding the early evolution of Titanosauriformes. Full preparation and description of this specimen, known from teeth, vertebrae and most of the appendicular skeleton of a single individual, recognises it as a distinct taxon: Vouivria damparisensis gen. et sp. nov. Phylogenetic analysis of a data matrix comprising 77 taxa (including all putative brachiosaurids) scored for 416 characters recovers a fairly well resolved Brachiosauridae. Vouivria is a basal brachiosaurid, confirming its status as the stratigraphically oldest known titanosauriform. Brachiosauridae consists of a paraphyletic array of Late Jurassic forms, with Europasaurus , Vouivria and Brachiosaurus recovered as successively more nested genera that lie outside of a clade comprising ( Giraffatitan + Sonorasaurus ) + ( Lusotitan + ( Cedarosaurus + Venenosaurus )). Abydosaurus forms an unresolved polytomy with the latter five taxa. The Early Cretaceous South American sauropod Padillasaurus was previously regarded as a brachiosaurid, but is here placed within Somphospondyli. A recent study contended that a number of characters used in a previous iteration of this data matrix are ‘biologically related’, and thus should be excluded from phylogenetic analysis. We demonstrate that almost all of these characters show variation between taxa, and implementation of sensitivity analyses, in which these characters are excluded, has no effect on tree topology or resolution. We argue that where there is morphological variation, this should be captured, rather than ignored. Unambiguous brachiosaurid remains are known only from the USA, western Europe and Africa, and the clade spanned the Late Jurassic through to the late Albian/early Cenomanian, with the last known occurrences all from the USA. Regardless of whether their absence from the Cretaceous of Europe, as well as other regions entirely, reflects regional extinctions and genuine absences, or sampling artefacts, brachiosaurids appear to have become globally extinct by the earliest Late Cretaceous.