Viewing an urban water system as a complex adaptive system provides new opportunities for analysis and avoids some critical simplifications. Taking this perspective, it is possible to explore the inter-related effects of changes to the system. This is particularly important in the developing world where donors providing aid aim to improve conditions but struggle to understand and quantify the systemic impacts of their actions. This is because an intervention aiming to improve condition may also have unintended and undesirable effects. To provide decision support, this paper describes an agent-based model of an urban water system, developed on the basis of ethnographic interviews, and subsequently evaluated by local stakeholders. The paper describes the model design as well as the results of scenarios. The model provides guidance on which system amendments may produce the best outcomes in terms of output variables, and on the basis of sense-checking and sensitivity analysis it is judged that model results are likely to give a good indication about possible real world outcomes. It is clear that no single strategy will solve all problems on its own, but that a combined strategy e with a strong focus on groundwater management and protection e is likely to be most successful.