BACKGROUND The methods currently used for measuring environmental tobacco smoke (ETS) exposure among small children all have their limitations. The aim of this study was to compare the results of questionnaire assessments of children's ETS exposure with cotinine measurements in urine and nicotine measurements in hair, a new method for estimating average ETS exposure. METHODS Questionnaire information on ETS exposure and a sample of hair were collected from 94 children aged 12-36 months. A urine sample for cotinine analysis was obtained from 72 of the children. RESULTS Nicotine was found in all hair samples and cotinine in all urine samples. Compared to children registered as unexposed by the questionnaire, hair nicotine levels were 12.4 times higher among children exposed to more than 10 DNC (daily number of cigarettes) (P < 0.001) and 3.6 times higher among children exposed to 1-10 DNC (P < 0.001). The median cotinine creatinine ratio (CCR) was 2.4 times higher among children exposed to more than 10 DNC compared to unexposed children (P < 0.001). No significant difference in median CCR was found between unexposed children and children exposed to 1-10 DNC. The correlation coefficient was 0.64 between children's hair nicotine levels and DNC, 0.50 between CCR and DNC and 0.56 between children's hair nicotine levels and CCR. CONCLUSION Nicotine measurement in hair is a practical and valid method for estimating average ETS exposure in children. An underreporting of ETS exposure was indicated.