Human exposure biomonitoring in the African Copperbelt (Katanga, southern D.R. Congo) revealed elevated cobalt (Co) exposure in the general population. This study was designed to identify the Co exposure routes for the non-occupationally exposed population in that area. The concentration of Co was measured in environmental and urine samples collected in urban and rural communities close to metal mining and/or refining plants, villages near a lake receiving effluents from metal refining plants, and control rural areas without industrial pollution. Drinking water, uncooked food items (maize flour, washed vegetables, fish and meat), indoor and outdoor dust samples were collected at each location. A food questionnaire was used to estimate dietary Co intake for adults and children. Geometric mean urine-Co (U-Co) concentrations were 4.5-fold (adults) and 6.6-fold (children) higher in the polluted than in the control area, with U-Co values being intermediate in the lakeside area. Average Co concentrations in environmental samples differed 6-40-fold between these areas. U-Co was positively correlated with most environmental Co concentrations, the highest correlations being found with Co in drinking water, vegetables and fruit. Estimated average total Co intake for adults was 63 (±42) μg/day in the control area, 94 (±55) μg/day in the lakeside villages and 570 (±100) μg Co/day in the polluted areas. U-Co was significantly related to modelled Co intake (R(2)=0.48, adults and R(2)=0.47, children; log-log relationship). Consumption of legumes, i.e. sweet potato leaves (polluted) and cereals+fish (lakeside) was the largest contributor to Co intake in adults, whereas dust ingestion appeared to contribute substantially in children in the polluted area. In conclusion, dietary Co is the main source of Co exposure in the polluted area and Co is efficiently transferred from soil and water in the human food chain.