The validity of stroke diagnosis in the National Hospital Discharge Register and the Register of Causes of Death was examined among 546 middle-aged men in Finland. The subjects were cases of cerebrovascular diseases of the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study and identified by record linkage to the registers. In all, 375 events with cerebrovascular disease as hospital discharge diagnosis and 218 events with cerebrovascular disease as the underlying cause of death were reviewed using specific criteria modified from the classifications of the National Survey of Stroke and the WHO MONICA Study. For hospital stroke diagnoses, there was agreement on diagnosis for all strokes in 90%, for subarachnoid hemorrhage in 79%, intracerebral hemorrhage in 82%, and cerebral infarction in 90%. The respective agreement rates for stroke as the underlying cause of death were 97%, 95%, 91%, and 92%. The data were insufficient for review in 1% and 3% of the stroke events, respectively. Age, observation year and trial supplementation with alphatocopherol or beta-carotene had no effect on validity. In conclusion, the validity of stroke diagnosis was good in registers of hospital diagnoses and causes of death justifying their use for endpoint assessment in epidemiological studies.