A retrospective cohort study was carried out in The Netherlands to investigate the potential carcinogenic effects in humans of occupational exposure to acrylonitrile (AN). The total study group consisted of 6803 workers "from eight chemical plants and one control plant" of whom 2842 had been exposed to AN between January 1, 1956 and July 1, 1979 for at least 6 months. All workers were employed by one of eight chemical companies. An extensive review of the available industrial hygiene data was conducted to assess the magnitude of past exposure to AN, occurrence of peak exposures, exposure to recognized potential human carcinogens, and respirator use. The total cohort was observed for mortality until January 1, 1988. In collaboration with the Central Bureau of Statistics, the causes of death were traced for the workers who died before 01-01-1988. In the exposed as well as in the nonexposed cohorts the total mortality was lower than expected, based on national mortality statistics. The observed cancer mortality in the exposed cohort was similar to the expected mortality. Specific analyses were carried out to investigate dose-response relationships and latency for total mortality and lung cancer mortality. Overall, no indications were found for a carcinogenic effect in this cohort of workers exposed to AN.