This article addresses the question ‘Does aid work?’ by asking ‘How do we know if it works?’ Despite substantial refinement in evaluation approaches, evaluation remains without any orthodoxy about how to assess effectiveness. The article examines the purposes of evaluation to discern unresolved tensions between accountability and an organizational learning approach. This is framed by the current vogue of Managing for Development Results, integral to the Paris Declaration on Aid Effectiveness in attaining the Millennium Development Goals. The article appraises the ‘paradigm war’ revealed in debates over methodology and the utility of the logical framework. It critiques limitations in development evaluation doctrine to highlight profound uncertainties of attribution and causality. The ‘evaluation gap’ between the rhetoric of donors and their evaluation practice, unsurprisingly, inhibits the evaluation of aid effectiveness.