At present, the micronucleus test is the most popular short-term assay for the in vivo detection of clastogens or spindle poisons. As micronuclei are rare events, it has been argued that the conventional microscopical analysis based on 500-2000 cells per animal does not provide enough sensitivity. In addition, the manual scoring of micronuclei is time-consuming and tedious. As an attempt to solve these problems, an automated method for the analysis of micronuclei in mouse bone marrow erythrocytes was developed using flow cytometry. Femoral bone marrow cells, from mice treated with known in vivo clastogens [mitomycin C (MMC): 0.5, 1 or 2 mg/kg body weight intraperitoneally; benzene: 1000, 2000 or 4000 mg/kg body weight orally] or vehicles for the clastogens (physiological saline, olive oil; for control animals), were fixed with 1% glutaraldehyde, suspended in 70% ethanol for storing, stained with 4',6-diamidino-2-phenylindole (0.5 microgram/ml) and analyzed (50,000 cells per sample) in an EPICS V flow cytometer using a Coherent 5W UV laser at 150-200 mV. The frequency of micronucleated erythrocytes (MNEs) was calculated from the cytometer, by a computer program involving model fitting to normal Gaussian distribution. Correlations between MNE frequencies obtained by manual microscopic observations (1000-2000 cells per animal) and by the flow cytometric measurements were good for both group results (means of 4-5 mice; linear correlation coefficient r = 0.89-0.998) and individual data (r = 0.82-0.96). Repeated experiments with MMC showed good reproducibility for the flow cytometric scoring of micronuclei.