Protection of groundwater resources is essential to ensure quality and sustainable use. However, predicting vulnerability to anthropogenic pollution can be difficult where data are limited. This is particularly true in the Sahel region of Africa, which has a rapidly growing population and increasing water demands. Here we use groundwater measurements of tritium (3H) with machine learning to create an aquifer vulnerability map (of the western Sahel), which forms an important basis for sustainable groundwater management. Modelling shows that arid areas with greater precipitation seasonality, higher permeability and deeper wells or water table generally have older groundwater and less vulnerability to pollution. The most important spatially continuous predictor was elevation. Groundwater vulnerability is based on recent recharge, implying a sensitivity also to a changing climate, for example, altered precipitation or evapotranspiration. This study showcases the efficacy of using tritium to assess aquifer vulnerability and the value of tritium analyses in groundwater, particularly towards improving their spatial and temporal resolution.