BACKGROUND: There is considerable discussion over the possible harm caused by fetal exposure to mercury, but evidence of such harm is contradictory at levels commonly found in populations with moderate intakes of fish. Further information is needed to inform debate and clarify policy recommendations. MATERIAL: Data were collected prospectively for the Avon Longitudinal Study of Parents and Children (ALSPAC). Whole blood taken in the first half of pregnancy was assayed for mercury. The outcomes were offspring behavioural assessments collected using the Strengths and Difficulties Questionnaire at seven time points between ages 4 and 16-17 years; five were completed by the mother and two by the teacher. Socioeconomic and biological confounders were first taken into account; further analyses added maternal blood selenium. Separate analyses compared the relationships between prenatal mercury levels and behaviour traits treated as continuous measures in women who ate fish with those who ate no fish in order to determine whether the relationships differed; the hypothesis was that fish consumption had benefits on the brain and masked any mercury effects. In order to prevent Type II errors, the P value for significance was set at 0.10. RESULTS: Prenatal mercury measurements and offspring behaviour results were available for between 2776 (at 47 months) to 1599 mother-child pairs (at 16-17 years). Even given a P value of 0.10, the number of significant results was no greater than expected apart from the relationships with peer problems at 4, 6 and 10-11 years where the relationships with prenatal mercury were negative (i.e. the greater the level of mercury the fewer the problems the child had with his/her peers). There were no significant differences between the associations with mercury found among the offspring of women who ate fish in pregnancy and those who did not, nor did adjustment for selenium make a difference. CONCLUSIONS: There were no adverse effects of maternal prenatal mercury levels on the behaviour of the offspring. A similar lack of relationship was found when the analyses were confined to those offspring whose mothers had eaten fish in pregnancy, and no consistent differences were found between the fish and non-fish eaters.