We have used the glycophorin A (GPA) in vivo somatic cell mutation assay to assess the genotoxic potential of styrene exposure in 47 reinforced plastics workers occupationally exposed to styrene and 47 unexposed controls matched for age, gender, and active smoking status. GPA variant erythrocyte frequencies (Vf), reflecting GPA allele loss (phi/N) and allele loss and duplication (N/N) somatic mutations arising in vivo in the erythroid progenitor cells of individuals of GPA M/N heterozygous genotype, were flow cytometrically determined in peripheral blood samples from these subjects. Measurements of styrene exposure of the workers at the time of blood sampling showed a mean 8-h time-weighted average (TWA8-h) styrene concentration of 155 mg/m3 (37 ppm) in the breathing zone. Mean urinary concentrations of the styrene metabolites mandelic acid (MA) and mandelic acid plus phenyl glyoxylic acid (MA+PGA) were 4.4 mmol/liter (after workshift) and 2.1 mmol/liter (next morning), respectively. Multivariate analysis of covariance on log-transformed GPA Vf data with models allowing adjustment for age, gender, smoking status, and styrene exposure showed that N/N Vf were nearly significantly increased among all of the exposed workers (adjusted geometric mean, 6.3 per million versus 5.0 in the controls; P = 0.058) and were statistically significantly elevated (adjusted geometric mean, 6.8 versus 5.0 in the controls; P = 0.036) among workers classified into a high-exposure group according to personal TWA8-h concentration of styrene in the breathing zone of > or = 85 mg/m3 (20 ppm; Finnish threshold limit value). Women in this high exposure group showed especially elevated N/N Vf (adjusted geometric mean 8.5 versus 5.3 in control women; P = 0.020); this elevation was also significant if urinary MA+PGA of > or = 1.2 mmol/liter was used as the basis of classification (adjusted geometric mean, 8.3; P = 0.030). The occupational exposure could not be shown to influence phi/N Vf. Cigarette smoking was associated with significantly elevated GPA Vf among active smokers (P = 0.042 for phi/N and P = 0.020 for N/N) and among active and ex-smokers combined (P = 0.014 for N/N). Its influence on phi/N Vf was especially clear among active smokers in the control group (P = 0.005). An effect of smoking, nearly statistically significant, was also observed for the phi/N Vf of control ex-smokers (P = 0.055) and of all active and ex-smokers combined (P = 0.050). Thus, the two characterized chemical exposures experienced by this group of workers and controls appear to produce differential effects on the two independent classes of GPA variants enumerated in the assay. This result suggests that the genotoxicity of these agents is mediated, at least in part, by different genetic mechanisms. Styrene exposure is associated with a specific increase in GPA N/N Vf; these allele loss and duplication variants reflect predominantly somatic recombination mechanisms in erythroid progenitor cells. Tobacco smoke exposure in active and ex-smokers is also associated not only with an increase in N/N Vf but also with an increase in phi/N Vf, reflecting the induction of GPA gene-inactivating mutations, including point mutations and deletions. This finding is consistent with a broad mechanistic spectrum of tobacco smoke genotoxicity associated with this complex mixture of chemical mutagens. Finally, there was no detectable effect of age on phi/N Vf; however, a highly significant (P = 0.0002) increase in N/N Vf with age, even after adjustment for other variables, was observed.