It remains unclear whether hyperglycaemia as measured by HbA(1c) is a significant risk factor for stroke in patients with type 2 diabetes mellitus (T2DM). A case-control study nested in a prospective cohort with 1 : 3 controls matched on age, gender, systolic blood pressure and low-density lipoprotein cholesterol (LDL-C) was conducted. The case group included 105 patients who developed incident stroke during 2.88 years (SD: 1.59) of follow-up of 4150 T2DM patients and 299 matched patients without incident stroke, used as the control group. Stratified Cox proportional hazard regression was used to obtain hazard ratio (HR). Median age was 71 years (IQR: 9.0 for the case and 10.0 for the control). HbA(1c) was significantly higher in the cases than in the controls (median 8.0% [IQR: 2.0] versus 7.2% [2.1], p < 0.0001). After controlling for smoker status, haematocrit, drug treatments and other covariates, 1% increase in HbA(1c) was associated with 1.49 (95% CI: 1.18-1.88, p = 0.0008) folds risk of occurrence of incident stroke. Patients with a history of coronary heart disease (CHD) were also at increased risk of stroke (HR: 8.25, 95% CI: 2.22-30.73, p = 0.0016). Smoker status and haematocrit were marginally significant predictors of incident stroke. Every adjusted month using lipid-lowering drugs was significantly associated with reduced risk of incident stroke (HR: 0.95, 95% CI: 0.90-0.99, p = 0.0199). Similar analysis using ACEI or ARB as a drug group was marginally significant (p = 0.0555). Chronic hyperglycaemia is a risk factor of stroke in Chinese patients with T2DM. (c) 2007 John Wiley & Sons, Ltd.