This study first examined the accuracy of death certificate diagnoses of 4,954 cases of cancer of the lung, liver, nasopharynx, and pleura/peritoneum, then compared usual occupation and industry based on case selection from the Metropolitan Detroit Cancer Surveillance System (MDCSS), a population-based cancer registry, with cases selected from death certificates for the above types of tumors to examine the effect of misclassification. Accuracy of death certificate cancer diagnoses ranged from 93.4% for lung cancer to 28.6% for malignancies of the pleura/peritoneum. The mix of usual occupation/industry titles obtained from death certificate cases and MDCSS cases was similar for lung cancer but not for malignancies of the pleura/peritoneum (35.7% of cases from the registry v 11.1% from death certificates for the automobile industry, P = .05). The effect of misclassification and utility of usual occupation/industry statements on death certificates is discussed.