In non-smokers, haemoglobin adducts from 3- and 4-aminobiphenyl have been reported to arise mainly from exposure to environmental tobacco smoke (ETS). Therefore, the impact of self-reported smoking (n = 27) and exposure of non-smokers to ETS (n = 78) on haemoglobin adducts was studied in pregnant women from Homburg, Germany. In addition to 3- and 4-aminobiphenyl, adducts from seven monocyclic aromatic amines (aniline, o -, m -, and p -toluidine, 2,4-dimethylaniline, 2-ethylaniline and o -anisidine) and the adduct from tobacco-specific nitrosamines (4-hydroxy-1-(3-pyridyl)1-butanone) were determined. Five of 78 self-reported non-smoking women had plasma cotinine levels and urinary cotinine/creatinine ratios indicative of active smoking. In the remaining 73 non-smokers cotinine/creatinine ratios correlated significantly with self reported exposure to ETS. However, none of the haemoglobin adducts increased with increasing exposure to ETS or increasing cotinine/creatinine ratios. Although significantly elevated in smoking compared with non-smoking women, the mean haemoglobin adduct levels formed by tobacco-specific nitrosamines (54 7 8 9 vs 26 7 4 1 fmol g-1, p < 0 001), 3-aminobiphenyl (3 0 0 5 vs 1 4 0 1 pg g-1, p < 0 001), 4-aminobiphenyl (27 9 3 4 vs 10 2 0 7 pg g-1, p < 0 001), o -toluidine (289 25 vs 237 65 pg g-1, p < 0 001), p -toluidine (315 32 vs 197 13 pg g-1; p < 0 001), 2,4-dimethylaniline (25 5 2 9 vs 18 6 1 6 pg g-1, p < 0 05), had considerable overlappings ranges indicating lack of specificity as biomarkers to tobacco smoke exposure. Exposure to other as yet unknown environmental sources appearsto be more significant than previously thought.