Exposure to arsenic through domestic burning arsenic-containing coal causes various tumors in a population of Guizhou, China. The glycophorin A (GPA) assay is a human mutation assay detecting somatic variation in erythrocytes expressing the MN blood type, and was used to assess genotoxicity of arsenic-exposed patients. Peripheral blood was collected from 18 adult healthy subjects and 40 arsenic-exposed patients in heparin-treated tubes. Erythrocytes were isolated, fixed in formalin and immuno-labeled with fluorescent antibodies against GPA, followed by flow cytometry analysis. Arsenic exposure increased the variant frequency (expressed as the number of variant red cells per 10(6) erythrocytes): NN, 3.7 in healthy subjects versus 21.2 in arsenic-exposed patients; N phi, 12.6 versus 33.1; MM, 13.1 versus 110; and M phi, 5.2 versus 20.3. The total GPA variant frequency was increased about five-fold (34.7 in healthy subjects versus 185 in arsenosis patients). Furthermore, the variant frequency was significantly higher in skin tumor-bearing patients: NN, 19.4 in arsenic-exposed non-tumor patients versus 31.5 in tumor-bearing patients; N phi, 29.5 versus 54.5; MM, 102 versus 159; M phi, 15.9 versus 45.1. Total GPA variant frequency in arsenic-exposed patients bearing skin tumors was significantly increased compared to patients without skin tumors (167 versus 290). The relationship between arsenic exposure history and GPA variant frequency was less evident. These data demonstrate that arsenic exposure is associated with mutations at the GPA locus, an effect exaggerated in patients bearing arsenic-induced skin tumors. The variant frequency of GPA could be a useful biomarker for arsenic exposure and arsenic carcinogenesis.