A MEDLINE search identified 12 published epidemiologic studies that have reported incidence or mortality experience among workers exposed to acrylonitrile. Many of the studies contain scanty descriptions of subject ascertainment, and most do not have good information on exposure assessment. Many also may have suffered from incomplete follow-up, as evinced by an overall deficit in the number of deaths observed, compared with the number expected from general population mortality rates. Such problems are not unique to studies on acrylonitrile, and to some extent they reflect the difficulties of conducting retrospective cohort studies. Despite these drawbacks, a simplified meta-analysis of the mortality experience reported for these cohorts revealed little evidence for carcinogenicity. Approximately the same number of cancer deaths was observed as was expected according to general population mortality rates (standardized mortality ratio 1.03, 90% confidence interval 0.92-1.15). The combined information from these studies is insufficient to support confidence about a lack of carcinogenicity at all sites. Nevertheless, despite the flaws in some of the individual studies, the summarized findings offer reassurance that workers exposed to acrylonitrile face no striking increases in mortality for all cancers or for respiratory cancer.