In many types of development project, direct provision of benefit to ‘the poor’ is a central goal. But how effectively is pro-poorness achieved? We report an independent ex-post evaluation of the pro-poorness of the World Bank-financed Programme d'assainissement des quartiers périurbains de Dakar (PAQPUD) sewerage project in Dakar, Senegal; we also review ex-post evaluations of previous donor-funded sewerage projects in African cities. We conclude that Dakar was a questionable location for major donor funding, given that this city's sanitation status is already much better than that of most African cities. If we accept the location, the Dakar PAQPUD project was more genuinely pro-poor in intention than most similar previous projects; however, many difficulties arose at implementation, and within the intervention areas, many of the poorest households did not benefit. In view of these results and our review findings, we argue that planners need to pay greater attention to household-level targeting: i.e. to ensuring that the poorest households will actually connect to the system. There is also a clear need for independent assessment of pro-poorness at ex-post evaluation. It is important to evaluate outcome through consultation not only with beneficiaries, but also with non-beneficiaries within the project's universe of intervention, and to investigate why non-beneficiaries have not benefited.