This study was undertaken to evaluate the quality of diagnoses of sudden death (SD) caused by ischemic heart disease (IHD) on death certificates. A random sample of 10% (100 cases) was drawn from all such deaths which occurred in 1970 among Oklahoma City residents. The medical records of each case were reviewed and the quality of the diagnosis was rated, by the use of predetermined standard criteria, as confirmed and valid or unconfirmed and invalid. Among 100 IHD deaths, SD constitute 45%. Of these SDs, 18% were unobserved, without information as to the manifestations of the attack or the interval from onset of symptoms to death. According to criteria of the American Heart Association and WHO Expert Committee, the diagnosis in this 18% was unconfirmed and invalid. Thus there is insufficient quantitative evidence to justify the use of SD, as found on death certificates, as an indication of frequency of ischemic heart disease.