BACKGROUND Cytogenetic changes associated with ethylene oxide (ETO) exposure at a worksite prompted a study of cancer incidence in that cohort. METHODS Cancer incidence through 31 December 1987 was ascertained in a cohort of 1132 individuals employed at the worksite at any time from 1 July 1974 through 30 September 1980, the period of potential exposure to ETO at the plant. The number of observed cancers was compared with that expected based on age- and sex-specific incidence rates reported by the National Cancer Institute's Surveillance Epidemiology and End Results Program. Standardized morbidity ratios (SMR) were calculated separately for regular and temporary employees. RESULTS Of the 28 cancers observed in the cohort, 12 were breast cancers. The SMR for breast cancer among regular female employees ranged from 2.55 (95% CI: 1.31-4.98, P = 0.02) to 1.70 (95% CI: 0.89-3.23, P = 0.09) depending on calendar year of follow-up, assumptions about completeness of follow-up, and the reference rates used. The excess of breast cancer over expected in regular female employees diminished over time. No statistically significant excess of breast cancer was noted for temporary female employees at any point during follow-up. No increase in cancer incidence was found over that expected for any cancer sites associated with ETO in previous studies--leukaemia, brain, pancreas and stomach. CONCLUSIONS Factors such as appropriateness of latency periods, length of follow-up and lack of a common histopathological type need to be considered in evaluating the excesses in observed breast cancer incidence, which diminished over time.