Incidental observations of carcinogenic activity indirectly associated with ethylene oxide (EO) have suggested a need for the evaluation of its carcinogenic potential. The formal studies to evaluate the carcinogenicity of EO are reviewed, and the weight of the evidence is assessed overall. Dunkelberg administered EO to female NMRI mice by subcutaneous injection, the route chosen to quantify the applied dose and to make possible the assessment of local tumor development, apart from tumors at remote sites. Application of the scheme to the data by Dunkelberg, suggests that EO is a substance "acting only indirectly" or at most, a "weak carcinogen". A chronic EO inhalation study was carried out by Snellings et al. If EO were a potent carcinogen specifically for the gastric mucosa, then one would expect that the gastric reaction product of EO and HCl, 2-chloroethanol, would also be carcinogenic.