Urbanisation and climate change are likely to induce more floods in African cities. Nonetheless, studies on public and private adaptation to floods that centre on the urban poor in Africa are scanty. Studies in this area largely reflect the structuralist conception of adaptation. This study departs from this top-down approach as it explores household and public adaptation to urban floods among the poor in Accra from an actor-oriented perspective. Specifically, the study objectives are to: a) analyse the causes of flooding in poor urban communities in Accra from various actor perspectives; b) understand the actions and challenges of actors involved in flood adaptation; and c) determine the correlates of household flood risk and private proactive adaptation choices among the poor in Accra. The study applied both exploratory and cross-sectional designs. Data collection methods under the exploratory design were literature review, in-depth interviews with key informants and focus group discussions in three communities, namely, Glefe, Mpoase and Agbogbloshie. A mini workshop for stakeholders in flood adaptation in Accra was organised to brainstorm on challenges within the network of actors. The study employed Kendall‟s Co-efficient of Concordance, network maps and content analyses of in-depth interviews as well as focus group discussions to achieve the first and second objectives. The cross-sectional aspect of the study involved structured interviews with 330 households selected through multi-stage sampling and using logistic and ordered probability regressions to analyse the results of the household survey to achieve objective three. The study found out that the level agreement on the perceived causes of flooding among actors involved flood adaptation in Accra was rather low. The differences in opinion were influenced by externalisation of blame and responsibility among actors as well as different actor interests. The challenges to public adaptation to urban floods in Accra are legal pluralism, strict adherence to organisational goals among formal institutions involved in flood adaptation and poor integration of local knowledge into formal flood abatement systems. There is also mistrust between local communities and the metropolitan level actors. x At the household level, the predictors of flood adaptation choices were tenancy status, home elevation, type of wall material, perceptions about future occurrence of floods, perceived adaptation cost, perceived adaptation efficacy and availability of bonding social capital. The study also found out that taking precautionary measures ahead of floods and living in sandcrete houses away from water bodies and at high elevations reduced household susceptibility to property damage or loss from urban floods. The study recommends streamlining power relations among institutions involved in flood adaptation and integrating informal actors into the formal flood adaptation structures at the metropolitan level. Awareness creation programmes should focus on zoning regulations, future occurrence of floods and construction materials/methods in flood zones. Finally, in-situ community upgrading, flood zone planning and enforcement of zoning regulations is also recommended to minimise exposure to flood risk in the study communities.