The last fifteen years have seen an intensive discussion on the effectiveness of aid. Part of the debate focused on the (often confusing and conflicting) evidence from growth regressions regarding the effect of aid on economic growth in recipient countries. A different discussion (probably of more interest to policy makers) concerned the implications of different forms of aid and different types of conditionality for the incentives for recipient governments to use the aid as intended by the donor. One of the most influential contributions to this discussion was the Assessing Aid report, published by the World Bank in 1998. It is often taken for granted that the lessons from this debate have by now been largely implemented. This is not the case. The debate is now focused on new concerns, notably on the fragmentation of aid and the lack of harmonization amongst donors. The Paris and Accra declarations bear witness to this development. This has left an unfinished agenda. In this paper we argue that there remains a gap between the rhetoric which suggests that the lessons of the debate on the failure of conditionality have been fully absorbed and the reality which is quite different.