OBJECTIVE Fetal exposure to tobacco constituents is a risk factor for negative birth outcomes. We aimed to determine the relationships between nicotine and cotinine concentrations in amniotic fluid and maternal saliva. METHODS As part of a therapeutic trial, 42 pregnant smokers agreed to sample amniotic fluid (8 samples from amniocentesis, 34 at birth). Their smoking characteristics were collected along with the newborns' birth outcomes. RESULTS The median concentrations [IQR] in amniotic fluid and saliva were 11 [7-31] and 38 [7-174] μg/L for nicotine and 72 [22-123] μg/L and 55 [17-109] μg/L for cotinine, respectively. Multivariate models showed that saliva cotinine concentration predicted amniotic fluid nicotine and cotinine concentrations (R2 = 0.398, p < 0.0001 and R2 = 0.708, p < 0.0001 respectively). Amniotic fluid nicotine or cotinine concentration was not associated with birth weight. In multivariate analysis, the time elapsed since the last cigarette was the only variable associated with increased birth weight (R2 = 0.237, p = 0.002). CONCLUSIONS Maternal saliva sampling for the determination of cotinine concentration is of interest to monitor fetal exposure to nicotine of any origin. Nevertheless, the time elapsed since the last cigarette was a better predictor of birth weight than the biomarkers' concentrations in amniotic fluid or maternal saliva.