We reviewed the reported epidemiological studies on cerebrovascular disease in the Taiwan area and discussed the following advantages and drawbacks of different approaches to data acquisition and analysis of the results. 1) Analysis of the data from vital statistics is readily available, and gives adequate information for secular trends, but the reliability of death certificate and the accuracy of subtypes of cerebrovascular disease must be taken into consideration. 2) A series of patients based on hospital registration can provide detailed data about the clinical manifestation and subtypes of stroke; whereas selection bias among hospitals may distort some clinical profile, and patients who have not visited the hospital are not included. 3) A community survey can obtain representative distribution of diseases in a community, but it takes much manpower and cost. The diagnosis of stroke is not very reliable because of limited usage of computed tomography and insufficient participation by qualified neurologists. 4) A community based hospital study has the advantage of combining a hospital registry and a community survey. It is difficult to conduct this type of study in the current medical environment in Taiwan since the referral system has not been well established. 5) A study of the survivalship of stroke can provide valuable information about the natural course of cerebrovascular disease.