Quantifying population exposure level to tobacco smoke is important for investigating its adverse effects on human health. We aimed to investigate the feasibility and application of using population hair concentrations of nicotine and cotinine to indicate their exposure level to tobacco smoke among pregnant women. Our study recruited 256 mothers who delivered healthy babies and collected their hair samples from scalp, of which 172 mothers were self-reported non-passive smokers and the other 84 mothers were self-reported passive smokers. We analyzed nicotine and cotinine concentrations of the hair section grown during the early pregnancy. The linear relationship between cotinine and nicotine was developed and validated by internal cross-validation method. Our results revealed that self-reported passive smokers had higher concentrations of nicotine [2.08 (1.00-4.46) ng/mg hair, i.e. median value (inter-quartile range)] and cotinine [0.063 (0.041-0.148) ng/mg hair] than non-passive smokers [1.35 (0.58-2.59) ng/mg hair of nicotine and 0.049 (0.022-0.087) ng/mg hair of cotinine, respectively]. There existed a linear regression model between hair cotinine and nicotine concentrations, i.e. [cotinine] = 0.024 × [nicotine]+0.0184 (R2 = 0.756) for this population. The internal cross-validation squared correlation coefficient slightly increased from 0.689 to 0.734 with the training subjects varying from 20% to 90%, suggesting that this regression model had high robustness and predictive accuracy. It was concluded that nicotine in maternal hair can evaluate the hair cotinine level and reflect maternal passive exposure level to ambient tobacco smoke with high sensitivity.