OBJECTIVE This study was designed to evaluate the relationship between occupational exposure to acrylonitrile and cancer mortality. METHODS Workers (18079 white men, 4293 white women, 2191 nonwhite men, and 897 nonwhite women) employed in acrylonitrile production or use in the 1950s through 1983 were followed through 1989 for vital status and cause of death. Exposure-response relationships were evaluated from quantitative estimates of historical exposures. Tobacco use was determined for a sample of workers to assess potential confounding. Mortality rates between the exposed and unexposed workers in the cohort were compared using the Poisson regression. RESULTS Analyses by cumulative, average, peak, intensity, duration, and lagged exposure revealed no elevated risk of cancers of the stomach, brain, breast, prostate or lymphatic and hematopoietic systems. Mortality from lung cancer was elevated for the highest quintile of cumulative exposure. When the decile categories were used, the relative risk did not continue to increase at higher levels. Adjustment for cigarette use reduced the risk for lung cancer only slightly. Separate analyses for wage and salaried workers, long-term and short-term workers, fiber and nonfiber plants, and individual plants revealed no clear exposure-response patterns. CONCLUSIONS The results indicate that exposure to acrylonitrile at the levels studied is not associated with an increased relative risk for most cancers of a priori interest. The excess of lung cancer in the highest quintile of cumulative exposure may indicate carcinogenic activity at the highest levels of exposure, but analyses of exposure-response do not provide strong or consistent evidence for a causal association.