Groundwater exploitation is rising in the Mekong Delta, Vietnam, potentially exacerbating arsenic contamination from natural sources. We investigate trends and controls on contamination patterns throughout the Delta's multi-aquifer system as observed in a spatially exhaustive data set of arsenic measured in >40,000 wells, 10.5% of which exceed the WHO drinking water standard for arsenic (10 μg/L). We relate strong trends in the distribution of contamination among well samples to explanatory variables derived from 3D ancillary physicochemical data sets using logistic regression models. Parsimonious models describe much of the observed variability in arsenic occurrence, which differs considerably between subsets of wells tapping shallow versus deeper aquifer groups. In the shallowest Holocene-Pleistocene aquifers, arsenic occurrence is best described by distance to the Mekong river channels and delta front, depth, and location within fault-bounded zones of the region. The same model, however, fails to explain observations in the deeper group of Pliocene-Miocene aquifers. Among these deeper units, arsenic occurrence is rare except among older wells in near-river, heavily pumped areas. Our analysis is the first to examine both natural and anthropogenically mediated contributions to the distribution of arsenic throughout the Mekong Delta's multi-aquifer system, with implications for management of similarly affected basins throughout Southeast Asia.