The alkaline single cell gel electrophoresis (Comet) assay was applied to study the occurrence of DNA damage in peripheral lymphocytes of human subjects with occupational exposure to low levels of benzene (twelve gasoline station attendants, with average benzene exposure of 0.3 mg/m3, 8 h TWA). The results obtained show a significant excess of DNA damage in lymphocytes of exposed workers, compared to matched unexposed controls (p = 0.028, Mann-Whitney U-test). Averaged tail moment values, based on 100 cells/individual, were 1.900 microns in the exposed and 0.936 micron in the unexposed group. In addition, exposed subjects showed a clearcut excess of heavily damaged cells, with tail moments > 90th percentile of the overall distribution (13.5 vs. 6.5%, p = 0.013, Mann-Whitney U-test). No correlation was found between the extent of DNA damage and the ages or smoking habits of the subjects. In order to assess the plausibility of the involvement of benzene in the results of the ex vivo study, further experiments were performed treating in vitro peripheral lymphocytes from unexposed donors with benzene metabolites hydroquinone, benzoquinone and benzenetriol. In these experiments, all benzene metabolites exerted a marked effect on resting lymphocytes, the lowest effective concentrations being below 1 microgram/ml. Conversely, far greater concentrations were required for the induction of significant DNA damage in parallel experiments with hydroquinone on mitogen stimulated lymphocytes. Addition of the DNA repair inhibitor cytosine arabinoside (Ara-C, 1-10 micrograms/ml) partially restored the sensitivity of stimulated cells to hydroquinone, an indication of the active processing of induced DNA lesions in growing cells. These results are discussed also in relation to the role of peripheral lymphocytes as target tissue in the biomonitoring of human exposure to genotoxic agents.