The aim of this study was to assess gasoline filling-station attendants' exposure to benzene and to determine which biological exposure index (BEI), trans,trans-muconic acid (t,t-MA) or S-phenylmercapturic acid (S-PMA), shows better correlation with environmental exposure. Exposure to benzene was measured using passive samplers (Radiello) attached to the collar of the overalls of subjects (n=33) just before the work-shift (approximately 8h); analysis was performed by GC-FID. S-PMA and t,t-MA were determined, respectively, by an immunochemiluminescent assay based on specific monoclonal antibodies and by HPLC-UV at 264 nm. Both methods of biological monitoring were performed on beginning and end-shift urine samples, and expected t,t-MA and S-PMA values were calculated. Smoking habits and life-style were ascertained by means of a questionnaire. Both environmental and biological monitoring data showed that benzene exposure for gasoline filling-station attendants was low when compared with the respective ACGIH limit values (means-benzene: 0.044 mg/m(3); t,t-MA: 171 microg/g creatinine; S-PMA: 2.7 microg/g creatinine). No significant correlation was found between exposure to benzene and t,t-MA or S-PMA excretion data. The use of expected values was also experimented for S-PMA and t,t-MA. This consists of calculating, on the basis of the known half-life of the benzene metabolite, the concentration of that metabolite that a worker should present at the end of the work-shift, the difference between this value and the value actually found is a measure of benzene exposure during work. The use of expected values in biological monitoring did not improve correlations. At these low benzene levels, environmental monitoring seems to be the best method of evaluating individual exposure. However, biological monitoring remains useful, as a mean of assessing group exposure.