Reduction of stroke burden requires preventive interventions targeted at important risk factors. This report presents the analysis of risk factors for ischaemic stroke from a representative cohort of middle aged men from South Wales and south-west England. Data on risk factors were collected through validated questionnaires and physical and clinical measurements. Details of possible cerebrovascular events were retrieved, classified into ischaemic, haemorrhagic and uncertain subtypes, and validated. The ratio of definite ischaemic to definite haemorrhagic strokes was calculated. This showed that the vast majority of strokes of unknown subtype were likely to ischaemic. After exclusion of known haemorrhagic strokes and subarrachnoid haemorrhages the remaining strokes were labelled ischaemic. Hazard ratios for possible risk factors were calculated for all ischaemic, and for fatal and non-fatal strokes. There were 293 ischaemic strokes. Statistically significant age-adjusted hazard ratios were: 1.50 (95% confidence interval 1.16-1.95) for being in a manual social class, 1.82 (1.24-2.67) if smoking >15 cigarettes/d at enrollment, 1.19 (1.13-1.24) and 1.23 (1.14-1.34) per 10 mmHg increase in systolic and diastolic blood pressure, respectively, 0.67 (0.46-0.96) for the top quintile high density lipoprotein-cholesterol:cholesterol ratio compared to the bottom quintile, 2.04 (1.40-2.99) for presence of angina, 3.90 (2.01-7.58) for presence of atrial fibrillation, and 3.35 (1.90-5.80) for presence of diabetes. Risk factors were more strongly associated with fatal than non-fatal strokes. Multivariate analyses revealed that, while there was some attenuation of the effect of social class, angina and elevated BP, the risks from atrial fibrillation and diabetes were increased.