We evaluated the mortality and cancer experience of employees of the chemical industry in the United States and western Europe, as reported in the peer-reviewed literature between 1966 and 1997 (>1 million workers and >15 million person-years). Cohort studies (N = 461) were grouped (N = 181) so that specific populations could be traced from the earliest to the most recently published report, and we extracted observed and expected numbers of cases for each of 35 mortality and 23 cancer incidence endpoints. We then generated standardized mortality ratios or standardized incidence ratios and 95% confidence intervals, and undertook meta-analyses of subcohorts (for example, gender, latency, or duration of employment), as well as the entire cohort. With few exceptions, the observed cause-specific mortality and site-specific cancer incidences were reassuring: overall, 10% fewer deaths were observed than expected. Fewer than expected deaths from all causes, cardiovascular disease, noncancer respiratory disease, cirrhosis of the liver, and external causes were observed, some or all of which may be attributed to a "healthy worker effect." Meta-analyses revealed weak to moderate excesses of lung and bladder cancers likely caused by occupational exposure to known human carcinogens. We also observed a 10-15% increase in lymphatic and hematopoietic cancers. Additional research is required to gain a more complete understanding of the potential role that methodology and environmental or occupational influences may play in these associations.