This paper quantifies the effect of water pollution due to oil spillage on local economic development outcomes in Nigeria. We assemble a geo-referenced panel of more than 13,000 oil spills occurrences recorded in the country between 2006 and 2019, and develop a hydrological model that traces contaminant transport over water networks, allowing spill exposure to extend beyond the point of discharge. We use arguably as good as random exposure within close range of spill sites to distinguish between directly treated locations, upstream locations, and downstream locations along watersheds and exploit this setup in a staggered difference-in-differences framework to estimate impacts on local socio-economic outcomes, including a novel proxy for extreme poverty obtained by combining high-resolution residential buildings data and nighttime lights. Relative to comparable cells, spill-exposed cells exhibit marked declines in nighttime lights, remotely-sensed extreme poverty, and the number of residents without electricity. Candidate mechanisms include environmental degradation, with annual declines in forest cover and vegetation health, and increased out-migration from affected locations. Dynamic event-study estimates show that these effects intensify from four to twelve years post-spill. We relate our remotely-sensed proxies to high-resolution survey data in order to estimate money-metric magnitudes of economic damages. Finally, we investigate the relationship between global oil price shocks and oil spillage intensity, in order to trace the complete causal chain from global commodity markets to local development outcomes.