Cancer morbidity was investigated in a cohort of 2,170 ethylene oxide (EO)-exposed workers from 2 plants producing disposable medical equipment. The subjects had been employed for at least 1 year during the periods 1970-1985 and 1964-1985, respectively. The exposure to EO was assessed for each of six job categories in the plants with respect to each calendar year, on which basis values for individual cumulative exposure to EO (ppm-years) were calculated. The levels of hydroxyethyl adducts to N-terminal valine (HOEtVal) in hemoglobin fitted well with the values estimated for airborne exposure to EO. No increased cancer incidence was found [standardized morbidity ratio (SMR), 0.78; 95% CI, 0.49-1.21)]. No leukemia was observed, but one case of non-Hodgkin's lymphoma, one case of myeloma, and one case of polycythemia vera were diagnosed as compared with two expected hematopoietic and lymphatic tumors (SMR, 1.54; 95% CI, 0.32-4.5). No stomach cancer was detected as compared with the 0.5 case expected. There were no significant exposure-response associations between estimates of exposure to EO and cancer morbidity.