BACKGROUND Methyl mercury (MeHg) and metallic Hg are well known as neurotoxic agents. Dental amalgam contributes significantly to elemental Hg vapour exposure in the general population. There is little information about Hg concentration in human amniotic fluid (AF) of pregnant women and its potential toxic effect on the fetuses. OBJECTIVE Primary to assess the relationship between the presence of detectable mercury (Hg) concentration in human AF, number and surface areas of amalgam fillings of pregnant women; secondary to analyse their obstetric history and perinatal complications. METHODS Seventy-two pregnant women took part in this prospective study. One dentist recorded the dental status, presence, number and surface areas of amalgam fillings. Total Hg concentration in AF was determined in digested samples using automatic cold vapour atomic absorption equipment. The detection limit of Hg in AF, determined from blank readings, was 0.08 ng/ml. To estimate the dependence of the explanatory variables (such as number and surface areas of amalgam fillings, fish consumption, presence of liver or neurological diseases and smoking habits) on mercury concentration several linear regression models were built up. Stepwise logistic regression procedures were running on total sample and on patients with at least one amalgam filling (Positive Filling group = PF). Principal component analysis (PCA) provided two factors, which explained for more the 60% of the variance among the variables. RESULTS The overall mean Hg concentration in AF among all patients was 0.37+/-0.49 ng/ml. Nineteen (26.4%) women had a Hg concentration <0.08 ng/ml (Hg negative group). In 53 (73.6%) patients, with a concentration > or = 0.08 ng/ml (Hg positive group), the mean value of Hg was 0.49+/-0.52 ng/ml. The average number of amalgam fillings was 2.26 +/- 3.19 in the Hg negative group and 5.32+/-3.03 in the Hg positive group (ANOVA one-way p=0.04). A dependence of mercury concentration on number of amalgam fillings (p=0.03), surface area of the amalgam fillings (p=0.04) and fish consumption (p=0.04) was observed but not at a significant level. In stepwise logistic procedure the number of amalgam fillings gave a contribution to the model (p=0.04), although null value was included in the confidence intervals. We observed no statistically significant differences (chi2 test) among the patients with a Hg concentration <0.08 ng/ml (n=19) and those with a concentration > or = 0.08 (n=53) with regard to obstetric history and perinatal complications. CONCLUSIONS Number and surface areas of amalgam fillings influenced positively Hg concentration in AF but not at a significant level. Moreover Hg levels detected in AF were low and no adverse outcomes were observed through pregnancies and in the newborns.