Background Small vessel disease is reported to be a more common cause of ischaemic stroke in people with diabetes than in others. However, population based studies have shown no difference between those with and those without diabetes in the subtypes of stroke. We determined the rates and predictors of risk of stroke and its subtypes in the Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) trial. Methods 9795 patients aged 50-75 years with type 2 diabetes were followed up for a median of 5 years. Annual rates were derived by the Kaplan-Meier method and independent predictors of risk by Cox proportional hazards regression analyses. Results The annual rate of stroke was 6.7 per 1000 person years; 82% were ischaemic and caused by small artery disease (36%), large artery disease (17%) and embolism from the heart (13%); 10% were haemorrhagic. Among the strongest baseline predictors of ischaemic or unknown stroke were age (60-65 years, HR 1.98; >65 years, HR 2.35) and a history of stroke or transient ischaemic attack (TIA) (HR 2.06). Other independent baseline predictors were male sex, smoking, history of hypertension, ischaemic heart disease, nephropathy, systolic blood pressure and blood low density lipoprotein (LDL) cholesterol, HbA1c and fibrinogen. A history of peripheral vascular disease, low high density lipoprotein, age and history of hypertension were associated with large artery ischaemic stroke. A history of diabetic retinopathy, LDL cholesterol, male sex, systolic blood pressure, smoking, diabetes duration and a history of stroke or TIA were associated with small artery ischaemic stroke. Conclusions Older people with a history of stroke were at highest risk of stroke, but the prognosis and prognostic factors of subtypes were heterogeneous. The results will help clinicians quantify the absolute risk of stroke and its subtypes for typical diabetes patients.