Gauthier (1986) coined a stem-based definition using birds and sauropodomorphs. Later authors used genera to replace one or both of these specifiers: Currie (1997) used Plateosaurus and Diplodocus instead of Sauropodomorpha; Padian et al. (1999) used Cetiosaurus instead of Sauropodomorpha; Sereno (1998) used Neornithes and Saltasaurus. Most recently, Holtz and Osmólska (2004) used Cetiosaurus oxoniensis and Passer domesticus as specifiers, incorrectly citing this definition as based on Padian and May (1993) and Maryanska et al. (2002) (neither of which define Theropoda).
Using Cetiosaurus is problematic given the fragmentary nature of the original type species. There is little gained in my opinion to give great weight to the first named species within a clade or interpretations of the intentions of nineteenth century systematists, when their actions and ideas may have been modified by a century of work and accumulated collections. The active definition is a first order revision of the defnition in Sereno (1998) that uses the deeply nested, well-founded sauropod species Saltasaurus loricatus.