Abstract Background There is a need to test the fetal programming theoretical framework in nutritional epidemiology. We evaluated whether maternal seafood intake during pregnancy was associated with 8-year-old attention outcomes after adjusting for previous child seafood intake and cognitive function. We also explored effect modification by several single nucleotide polymorphisms (SNPs) related with polyunsaturated fatty acid (PUFA) metabolism. Methods Our final analyses included 1644 mother-child pairs from the prospective INMA (INfancia y Medio Ambiente) cohort study (Spain, recruitment between 2003 and 2008). We used food frequency questionnaires to assess prenatal and postnatal seafood consumption of the mother-child pairs. We evaluated attention function of the children through the computer-based Attention Network Test (ANT) and we used the number of omission errors and the hit reaction time standard error (HRT-SE). Parents reported child attention deficit hyperactivity disorder (ADHD) symptoms using the Revised Conners’ Parent Rating Scale Short Form (CPRS-R: S). We measured seven candidate SNPs in a subsample of 845 children. We estimated associations using regression models, adjusting for family characteristics, child seafood intake and cognitive functions at early ages, and to explore SNP effect modifications. Results Higher total seafood intake during early pregnancy was associated with a reduction of child ANT omission errors, 5th quintile (median = 854 g/week) vs 1st quintile (median = 195 g/week), incidence risk ratio (IRR) 0.76; 95% CI = 0.61, 0.94. Similar results were observed after adjusting the models for child seafood intake and previous cognitive status. Lean, large and small fatty fish showed similar results, and generally similar but less robust associations were observed with the other attention outcomes. Shellfish and canned tuna showed weaker associations. The association patterns were weaker in late pregnancy and null in child seafood consumption. Child rs1260326 (glucokinase regulator, GCKR) and child/maternal rs2281591 (fatty acid elongase 2, ELOVL2) polymorphisms showed nominal P-value for interactions <0.10 between total seafood intake and ANT outcomes. Conclusions After adjusting for previous child cognitive functions and child seafood intake, high pregnancy consumption (total, lean, small and large fatty fish) was independently associated with improvements of some 8-year-old attention outcomes. Genetic effect modification analyses suggest PUFA intake from seafood as a potential biological mechanism of such association.