Men assigned to units producing ethylene oxide by the chlorohydrin or direct oxidation processes and to other departments using ethylene oxide in two chemical plants were followed up for mortality from 1940 to 1988 (n = 1896). Based on findings from a previous study of these workers to the end of 1978, which identified confounding exposures, workers assigned to one unit with low ethylene oxide exposure potential were excluded (n = 278). Average duration of exposure was over five years and average follow up was 27 years, with all subjects at least 10 years from first exposure. The data did not support associations of ethylene oxide with all cancer types combined, leukaemia, non-Hodgkin's lymphoma, or brain, pancreatic, or stomach cancers. There were also no duration-response trends. The standardised mortality ratio (SMR) for total cancer was 86 (95% confidence interval 71-104) and did not increase for those hired the earliest and with long duration assignments. The results of this 10 year update and those of other recent studies of ethylene oxide workers do not confirm findings from animal studies and are not consistent with the earliest results reported among ethylene oxide workers.