The 1960 Indus Waters Treaty has a reputation as a rare instance of cooperation between an otherwise belligerent India and Pakistan. Based on archival research in South Asia and the US, this paper argues that the treaty failed to resolve broader tensions, but did establish a precedent by defining a river basin as a unitary object of development (and subject of negotiation). The idea of organizing cooperative development across a whole river reappeared in American proposals for resolving an India-Pakistan dispute over the Ganges and Brahmaputra rivers in divided Bengal (East Pakistan, now modern Bangladesh) during the later 1960s and 1970s. Such proposals, however, met with outright hostility from Indian politicians, bureaucrats, and press. The spectacular failure of river-basin scale cooperation in eastern South Asia contrasted with its relative success in the Indus Basin. The story of Ganges-Brahmaputra negotiations – or, more usually, the reasons for the lack of them – demonstrates the historical and geographical peculiarity of the original Indus treaty. The paper's broad conclusion is that schemes geared towards encouraging broader cooperation in South Asia rather than solving specific and clearly-defined sets of problems, tend to decontextualize the 'success' that they wish to recreate.