Titanosauria has been defined both as a node- and stem-based taxon. The active definition is stem-based, although this was not the form of the first definition.
Bonaparte and Coria (1993) coined Titanosauria but did not provide a definition. The first definition, a node-based definition based on Andesaurus delgadoi, was provided by Salgado et al. (1997). Wilson and Upchurch (2003:154) recently advocated use of the node-based definition of Salgado et al. (1997) for two reasons, namely it (1) captured the intent of Bonaparte and Coria (1993) and (2) has priority. Bonaparte and Coria (1993), however, only listed included taxa. They did not specify a definition, and thus there is no way to know whether they would have included newfound basalmost taxa. There is no issue with the second point. Salgado et al. (1997) were first to erect a definition, which like all others in their paper was node-based. But is a node-based definition the most effective for a clade with an increasing number of basal singleton taxa, some of which are based on very incomplete type speccimens?
Sereno (1998) and Wilson and Sereno (1998) proposed a stem-based defintion because the type material of Andesaurus delgadoi is poorly known other taxa considered basal titanosaurians (e.g. Chubutisaurus, Salgado et al. 1997) may eventually reside in a more basal position. Recently Salgado (2003) has also adopted a stem-based definition for Titanosauria proposed by Sereno (1998) and Wilson and Sereno (1998).
Thus the active definition is a first-order revision of that in Wilson and Sereno (1998) that includes two additional external specifiers (Euhelopus zdanskyi, Brachiosaurus brancai) for stability of taxonomic content in the face of potential phylogenetic change among immediate outgroups.