The DNA adduct levels in total white blood cells (WBC) and lymphocytes (LYM) isolated from the blood of the same individuals were evaluated using the 32P-postlabelling assay for bulky aromatic adducts. In this study, 68 male coke oven workers and 56 machines workers as a matched control were enrolled. Personal monitors were used to evaluate exposure to eight carcinogenic PAHs, including B[alpha]P, during an 8-h working shift. The exposure among coke even workers ranged widely from 0.6 to 547 micrograms/m3 and from 2 to 62,107 ng/m3, for carcinogenic PAHs and B[alpha]P, respectively. The respective values in controls were from 0.07-1.64 microgram/m3 and from 1-63 ng/m3. A significant correlation between WBC- and LYM-DNA adduct levels was found (r = 0.591, P < 0.001). DNA adduct levels in both WBC and LYM were significantly elevated in coke oven workers as compared with controls, but adduct levels were generally low (WBC: medians 2.61 vs. 1.83 LYM: 2.47 vs. 1.65 adducts/10(8) nucleotides). LYM-DNA adduct levels were significantly higher for smokers as compared with nonsmokers in both the exposed and control groups. No such differences in WBC-DNA adduct levels were observed. Positive significant correlations were found at the individual level between DNA adducts in both cell types and carcinogenic PAHs and/or B[alpha]P in the inhaled air (r = 0.38-0.45, P < 0.001). A significant correlation at the individual level between LYM-DNA adducts and urinary cotinine was also observed (r = 0.37, P < 0.001). No differences in DNA adduct levels could be attributed to GSTM1 or NAT2 genotype in either group. Nor was there any clear association of DNA adduct levels with combined GSTM1/NAT2 genotypes. The effect of personal exposure to carcinogenic PAHs on DNA adduct levels in both cell types was also investigated using a logistic regression model with adjustment for possible modulating effect of confounders (smoking, GSTM1, NAT2, age, plasma levels of vitamins A and E, body mass index and diet). The results showed that coke oven workers had a significantly (P < 0.05) increased adjusted Odds Ratio (OR = 4.2 and 3.9 for WBC and LYM-DNA adducts) for occurrence of higher DNA adduct levels as compared to controls. The results also showed that the relative risk of an increased prevalence of 'abnormal' values of DNA adduct levels was exposure-dose related. The influence of confounding variables was found not to be significant in this study of relatively limited size. In spite of this, the results suggest that the DNA adduct levels in LYM seem to be affected by smoking (OR = 1.8 for smokers) and are modulated by the influence of NAT2 genotypes (OR = 1.6 for slow acetylators). Our findings indicate that both cell types are generally suitable to monitor occupational exposure to PAHs, and the results suggest that coke oven workers, smoking individuals and slow acetylators sustain more genetic damage in their LYM-DNA from exposure to carcinogenic PAHs than individuals without these actors.