OBJECTIVE--To evaluate the relations between environmental benzene concentrations and various biomarkers of exposure to benzene. METHODS--Analyses were carried out on environmental air, unmetabolised benzene in urine, trans, trans-muconic acid (ttMA), and three major phenolic metabolites of benzene; catechol, hydroquinone, and phenol, in two field studies on 64 workers exposed to benzene concentrations from 0.12 to 68 ppm, the time weighted average (TWA). Forty nonexposed subjects were also investigated. RESULTS--Among the five urinary biomarkers studied, ttMA correlated best with environmental benzene concentration (correlation coefficient, r = 0.87). When urinary phenolic metabolites were compared with environmental benzene, hydroquinone correlated best with benzene in air. No correlation was found between unmetabolised benzene in urine and environmental benzene concentrations. The correlation coefficients for environmental benzene and end of shift catechol, hydroquinone, and phenol were 0.30, 0.70, and 0.66, respectively. Detailed analysis, however, suggests that urinary phenol was not a specific biomarker for exposure below 5 ppm. In contrast, ttMA and hydroquinone seemed to be specific and sensitive even at concentrations of below 1 ppm. Although unmetabolised benzene in urine showed good correlation with atmospheric benzene (r = 0.50, P < 0.05), data were insufficient to suggest that it is a useful biomarker for exposure to low concentrations of benzene. The results from the present study also showed that both ttMA and hydroquinone were able to differentiate the background level found in subjects not occupationally exposed and those exposed to less than 1 ppm of benzene. This suggests that these two biomarkers are useful indices for monitoring low concentrations of benzene. Furthermore, these two metabolites are known to be involved in bone marrow leukaemogenesis, their applications in biological monitoring could thus be important in risk assessment. CONCLUSION--The good correlations between ttMA, hydroquinone, and atmospheric benzene, even at concentrations of less than 1 ppm, suggest that they are sensitive and specific biomarkers for benzene exposure.