OBJECTIVES: This study uses 5-year population Type II diabetes mellitus data to determine the risk of stroke in Taiwan, taking into consideration the age and gender of patients, as well as medication pattern. METHODS: The 2,123,104 diabetic patients data used are from the Taiwan National Health Insurance Research Database covering the period from 1999 to 2004. Newly-diagnosed diabetic patient was defined as had 3 or more visits or at least a diabetic code during admission or emergency visit in 1999 with a diagnosis of (ICD-9-CM code 250) and who did not have any diabetic code in 1998, a total of 681,188 were identified from the database. 45,949 sample patients were identified from the database by a principal diagnosis of stoke (ICD-9-CM code 430-438) during next 5 years. The analyses were descriptive in nature, including numbers and percentages for the categorical variables. Multiple logistic regression was used to compute the relative risk for the associations between different medication rate, health service type, physicians, diabetic patients and clinical complications. RESULTS: We find that for the period under examination, there was statistic significant increase in diabetic stroke incidence with male, age, medication rate, hospital level visit rate, same physician visit rate and bed sore, neuropathy complications. CONCLUSIONS: Our finding of significant risk of stroke in diabetic patients of different medication pattern may be explained by severity of disease and important of management of diabetic patients.