Studies of family history of cancer have contributed to knowledge about susceptibility to cancer and cancer genetics. The gold standard is to validate reported history of cancer by review of medical records and death certificates. Difficulties encountered in validating the history of cancer include the following: inaccuracy of diagnoses from death certificates when compared with medical records or autopsy data (7-5); coding inconsistencies, such as the report of nonspecific site of cancer on death certificates when compared with medical records (4); and inability to obtain records (5). Retrieval of medical records and accuracy of data obtained are higher if the history is given by firstdegree relatives (5, 6).

All patients diagnosed with tumors classified as Ewing's sarcomas and treated at the National Cancer Institute between 1965 and 1992 as well as their family members were enrolled in a genetic epidemiology study of Ewing's sarcoma. Approval of the institutional review board was obtained (7). The informant (ie, living proband or next of kin), after providing written consent, was asked to report diagnoses of cancers in the proband's first-and second-degree relatives. We present results of validation of reported cancer diagnoses in a subset of 122 of 320 deceased family members. Confirmation of reported cancer diagnoses through medical documentation (death certificates and medical records) was attempted systematically for diagnoses of melanoma, cancers of the brain, stomach, and lung, and for all cancers for which the accuracy of diagnosis was questioned. The methods have been fully described (7). Requests for death certificates were sent to vital records departments of the 30 states where case subjects resided. The mean waiting period was 2.5 months