The validity of death certificate diagnosis of out-of-hospital sudden cardiac death (OOH-SCD) was studied among 108,676 30- to 74-year-old residents in 5 Minnesota communities using 6-year mortality data (1985 to 1990). Among 4,244 total deaths, location of death was listed on the certificate as out of hospital in 2,035 cases. Of those, 911 were judged not to have OOH-SCD because they had actually been admitted to the hospital or were noncardiovascular deaths. Among the remaining 1,124, 254 were diagnosed as OOH-SCD using a thorough, physician-based procedure that used clinical records, autopsy reports, and an informant (next-of-kin) interview. We used only death certificate information to define OOH-SCD simply and inexpensively as ICD-9 code 427.5 (cardiac arrest) plus location of death listed as out-of-hospital. Compared with the physician diagnosis, sensitivity was only 24%, whereas specificity was 85%. When the definition of OOH-SCD was expanded to include ICD codes 410-414 (acute myocardial infarction and chronic coronary artery disease), sensitivity improved to 87%, whereas specificity became 66%. However, even with this higher sensitivity and specificity, only 27% of the cases labeled OOH-SCD by death certificate agreed with the physician diagnosis. Death certificate diagnosis of OOH-SCD included many erroneous cases, and may not have been suitable for study of etiologic factors, such as cardiac dysrhythmias. Death certificate diagnosis may be useful to assess population time trends in OOH-SCD, provided that misclassification (false-positive rate) remains constant over time.