AbstractObjectivesTo examine the association between fish consumption and stroke risk.DesignProspective population cohort study.SettingNorfolk, UK cohort of the European Prospective Investigation into Cancer (EPIC–Norfolk).SubjectsSubjects were 24 312 men and women aged 40–79 years who had no previous history of stroke at baseline.MethodsFish consumption was assessed using a food-frequency questionnaire at baseline in 1993–1997 and stroke incidence ascertained to 2004.ResultsA total of 421 incident strokes were identified (mean follow-up=8.5 years, total person-years=209 238). There were no significant relationships between total fish, shellfish or fish roe consumption and risk of stroke in men and women after adjusting for age, systolic blood pressure, body mass index, smoking, cholesterol, diabetes, physical activity, alcohol consumption, fish oil supplement use and total energy intake using Cox regression analyses. Oily fish consumption was significantly lower in women who subsequently had a stroke (odds ratio (OR) for consumers vs. non-consumers=0.69, 95% confidence interval (CI) 0.51–0.94,P=0.02). The trend in men was similar but not significant (OR for consumers vs. non-consumers=0.88, 95% CI 0.65–1.19,P=0.41).ConclusionsThere was no consistent relationship between fish consumption and stroke in this British population. Inconsistencies in the observed health effects of fish consumption in different populations may reflect different patterns and type of fish consumed and preparation methods.