Sister-chromatid exchanges (SECs) and chromosomal aberrations (CAs) were detected in the peripheral lymphocytes of 41 sanitary workers exposed to ethylene oxide (EO) in the sterilizing units of 8 hospitals in the Venice Region. The first group (19 workers) was exposed to 10.7 +/- 4.9 ppm EO, expressed as the time-weighted average concentration for an 8-h working day (TWA/8 h conc.), and the second group (22 workers) to 0.35 +/- 0.12 ppm. Each exposed worker was paired with a control of similar age and smoking habits. A highly significant (P less than 0.001) increase in the mean frequency of SCEs was found in the higher exposure group, 14 (74%) exposed subjects having significantly increased levels of SCEs compared to their matched controls. In the lower exposure group, the increase in mean frequency of SCEs was lower, though still significant (P less than 0.05): 7 (33%) exposed subjects had higher and 1 (5%) had a lower SCE level than the matched controls. From the first group, 10 subjects, 7 of whom had increased SCE levels, were reanalysed 12-18 months after their exposure had been lowered or interrupted: in only 2 of them the SCE level was significantly decreased. A statistically significant correlation between SCE frequency and level of EO exposure (TWA/8 h conc.), as well as a multiple correlation between SCE level and EO exposure, smoking and age were found. However, no interaction could be detected between EO exposure and smoking in the induction of SCEs. In controls, SCE frequency was correlated with smoking and age. In the higher exposure group, the number of both chromatid- and chromosome-type aberrations, independent of gaps, was significantly increased, whereas in the lower exposure group only the frequency of chromosome-type aberrations, excluding gaps, was statistically higher than in controls. The level of CAs remained to a great extent unchanged in the 10 subjects re-examined at a later stage after lowering or halting exposure. Taking the group as a whole, the frequency of cells with total CAs was found to be weakly (P = 0.05) correlated with EO exposure, and was not correlated with smoking, age or SCE frequency.