Acrylamide (AA) is a food-borne toxicant suspected to be carcinogenic to humans. It is formed in the heating process of starch-containing food. Currently, there is a great discussion about the possible human health risks connected with the dietary uptake of acrylamide. Haemoglobin adducts of acrylamide and its oxidative metabolite glycidamide are both markers of biochemical effect. However, because glycidamide has a higher carcinogenic potency than acrylamide itself, the glycidamide adduct might mirror the genotoxicity better than acrylamide adducts. In order to gain more information about the human metabolism of acrylamide, we investigated a small group of persons for the effective internal doses of acrylamide and glycidamide using haemoglobin adducts as parameters of biochemical effect. The collective was subdivided into non-smokers (n=13) and smokers (n=16) by determining the smoking-specific acrylonitrile haemoglobin adduct (N-cyanoethylvaline, CEV). The mean values for the adducts of acrylamide (N-2-carbamoylethylvaline, AAVal) and glycidamide (N-(R,S)-2-hydroxy-2-carbamoylethylvaline, GAVal) in nonsmokers was 19 pmol/g globin AAVal (range 7-31 pmol/g globin) and 17 pmol/g globin GAVal (range 9-23 pmol/g globin). For smokers mean levels of AAVal were 80 pmol/g globin (range: 25-199 pmol/g globin) and those of GAVal were 53 pmol/g globin (range: 22-119 pmol/g globin). Metabolism to glycidamide turned out to be significantly more effective in non-smokers than in the higher exposed smokers. Compared with studies in rats, the metabolic conversion of acrylamide to glycidamide as measured by haemoglobin adducts seems to occur to a similar extent in humans as in rats. Risk estimations on acrylamide based on experimental data obtained in rats obviously did not overestimate the cancer risk for the general population. Furthermore, our results might indicate that the dose-response curve for acrylamide is not linear. This would be in line with the results of animal experiments on rodents.