To prospectively assess the associations between lean fish, fatty fish and total fish intakes and risk of stroke in the Spanish cohort of the European Prospective Investigation into Cancer and Nutrition (EPIC-Spain).

Fish intake was estimated from a validated dietary questionnaire. Cox proportional hazards regression models were used to assess the association between the intakes of lean fish, fatty fish and total fish and stroke risk. Models were run separately for men and women.

Five Spanish regions (Asturias, San Sebastian, Navarra, Granada and Murcia).

Individuals (n 41 020; 15 490 men and 25 530 women) aged 20–69 years, recruited from 1992 to 1996 and followed-up until December 2008 (December 2006 in the case of Asturias). Only participants with definite incident stroke were considered as cases.

During a mean follow-up of 13·8 years, 674 strokes were identified and subsequently validated by record linkage with hospital discharge databases, primary-care records and regional mortality registries, comprising 531 ischaemic, seventy-nine haemorrhagic, forty-two subarachnoid and twenty-two unspecific strokes. After multiple adjustments, no significant associations were observed between lean fish, fatty fish and total fish consumption and the risk of stroke in men or women. In men, results revealed a non-significant trend towards an inverse association between lean fish (hazard ratio=0·84; 95 % CI 0·55, 1·29, Ptrend=0·06) and total fish consumption (hazard ratio=0·77; 95 % CI 0·51, 1·16, Ptrend=0·06) and risk of total stroke.

In the EPIC-Spain cohort, no association was found between lean fish, fatty fish and total fish consumption and risk of stroke.