This study investigates whether consumption of fish during pregnancy may prolong gestation or increase fetal growth. From 1984 to 1987, 11 980 pregnant Danish women filled out a questionnaire while they were in the 36th week of gestation; this was 83% of all such women living in two geographically defined areas. The women were divided into four strata dependent on having consumed a fish meal 0, 1-2, 3-4, or 5+ times during the previous month. In nonsmokers, a one level increase in fish consumption was, after adjusting for 19 other variables in a multiple regression model, found to be associated with an increase of 11 g in placental weight, 0.08 cm in head circumference and 16 g in birthweight (95% confidence intervals (Cl) 5 to 17 g (p = 0.0002), 0.02 to 0.14 cm (p = 0.02), and -2 to 34 g (p = 0.09), respectively); no associations with birth length or gestational age could be detected (95% Cl -0.07 to 0.11 cm (p = 0.7) and -0.82 to 0.05 days (p = 0.2), respectively, per level change in fish consumption). None of the associations seen in non-smokers were seen in smokers. We postulate that the associations seen in non-smokers could be due to marine n-3 fatty acids improving placental blood flow by increasing the ratio of prostacyclins to thromboxane. Smoking may possibly interfere with this mechanism via nicotine's inhibitory effect on platelet thromboxane production.