The validity and coverage of hospital records of stroke and myocardial infarction were evaluated in a Finnish epidemiological cohort consisting of 57000 men and women. Information on their hospitalizations after the baseline examination was obtained by record linkage to the National Hospital Discharge Register. As compared to data based on re-examination of survivors after five years, the discharge register covered 78.2% of hospital treatments. A specific diagnosis code had been recorded for 81.7% of the definite strokes and for 84.7% of the definite myocardial infarctions. The 5-year incidence rates based on both hospital records and deaths were clearly lower than the rates previously reported from local registers of stroke and myocardial infarction in Finland. Although hospital discharge information underestimates morbidity in the population, the data are sufficiently valid for many epidemiological purposes, e.g. research on predictors of diseases.