The hypothesis was tested that the absorption of labile Al in rats will increase when the Al-binding capacity of food components in the stomach is saturated. Male rats were exposed to 0, 10, 50, or 500 mg labile Al/L in acidic drinking water (pH 3) for 9 wk. The results show that labile Al in drinking water is complexed by feed constituents in the stomach of the rat in vivo, thus causing a nondetectable absorption of Al at 10 mg Al/L. An increased absorption of Al at 50 and 500 mg Al/L was associated with a saturation of the Al-binding capacity of feed components in the lumen of the stomach, causing the appearance of labile Al. Thus, the presence of labile Al in drinking water does not necessarily result in a high Al absorption when the water is ingested, since the bioavailability of labile Al is dependent both on the amount and composition of Al-binding components present in the gastrointestinal tract at the time of ingestion of the water. It is thus not possible to predict the body burden of Al in humans just by measuring the Al concentrations in drinking water. Even a further refining of the exposure measurement to include speciation of Al in the water may not markedly improve the prediction of the Al body burden. Future epidemiological studies must therefore be based on actual measurements of Al concentration in tissues or fluids from the study subjects.