A more detailed analysis based on facility, SES, length of employment, and period trend revealed substantial differences in the death rates among workers at the four Oak Ridge sites for white males and females (see Table III). Estimates of the internally adjusted log SMRs (facility effect estimates) show that the TEC, K-25, and multiple facility workers have higher death rates than similar workers employed only at X-10 or Y-12. The LRT statistics indicate that the differences among the facilities are primarily due to noncancer causes of death. However, further analyses of selected cancer causes (see Table IV) shows that there are large differences among the four facilities for lung cancer, leukemia, and other lymphatic cancer. Monthly paid workers had substantially lower mortality than nonmonthly workers; this is in agreement with other studies showing socioeconomic differentials in mortality. For white males, paycode differentials in mortality were greatest for respiratory diseases in general and emphysema in particular. These observations are consistent with studies showing socioeconomic differentials in smoking. Excess mortality was also observed for workers employed for less than one year, and among white males the relative risk was greatest for external causes of death. Over calendar time SMRs tended to increase. For white males, all cause SMRs increased at an average rate of 0.49L% per year, with values of 0.58L% for all cancer, 0.59L% for diseases of the circulatory system, and 0.97L% for respiratory diseases. Among the causes of death listed in Table III for white males, only emphysema showed decreasing SMRs over time. The magnitude of changes in SMRs with calendar time and differences in trends for causes of death of interest, including cancer and smoking-related diseases, suggests the importance of continued follow-up of these populations.