The mortality experience of 88,000 Union Carbide Corporation employees from 1974 to 1983 is presented using a population-based surveillance system. The study included many long-term employees, with most deaths contributed by those retired or terminated. The total population exhibited 30% lower mortality overall and 10% lower cancer mortality, as compared with the general U.S. population. Excesses of benign neoplasms and malignant melanoma of the skin were observed in both hourly and salaried males. Mortality rates for lymphosarcoma and reticulosarcoma were significantly elevated due to higher rates among hourly male employees and a cluster in one location. This same location also exhibited an excess of liver cancer associated with vinyl chloride operations. There were no other significant excesses in the hourly male workers and fewer deaths than expected due to brain cancer, respiratory cancer, and nonmalignant respiratory diseases. Salaried, and particularly hourly, women experienced favorable mortality, although for the women, time since hire was relatively short. Location-specific findings were similar to what had been observed in the company's previously conducted cohort studies. Future value lies in the development of a database that will have greater power to address possible effects of past exposures and outcomes related to more recent lower level exposures.