Studies of the beneficial role of fish consumption in the prevention of CVD are not consistent in their findings, particularly those studies that focus on the risk of stroke. The aim of the present study is to investigate the relationship between the consumption of different types of fish and the subsequent incidence of cerebrovascular disease (CVA). We prospectively evaluated the association between consumption of different types of fish and CVA in 3958 men and women aged 40–79 years who were free of heart disease and had participated in a health examination survey from 1967 to 1972. A total of 659 incident cases of CVA occurred during a follow-up until the end of 1994. A dietary history interview method provided data on habitual consumption of fish and other foods over the preceding year at baseline. Total fish intake did not predict CVA, but consumption of salted fish suggested an increased risk of intracerebral haemorrhage. The relative risk of intracerebral haemorrhage between the highest tertile of salted fish consumption and non-consumers was 1·98 (95 % CI 1·02, 3·84; P for trend = 0·06) after adjustment for age, sex, energy intake, smoking, BMI, physical activity, geographic area, occupation, diabetes, use of post-menopausal hormones, serum cholesterol, hypertension, and consumptions of butter, vegetables, fruits and berries. The relationship between fish consumption and stroke risk is not straightforward. How the fish is prepared for consumption may play an important role, affecting the association.