Phenol is not reliable as a biomarker for exposure to benzene at concentrations below 5 ppm (8-hr time-weighted average [TWA]). S-Phenylmercapturic acid (S-PMA) and trans-trans-muconic acid (tt-MA), two minor urinary metabolites of benzene, have been proposed as biomarkers for low-level exposures. The aim of this study was to compare their suitability as biomarkers. S-PMA and tt-MA were determined in 434 urine samples collected from 188 workers in various settings in the petrochemical industry and from 52 control workers with no occupational exposure to benzene. Benzene concentrations in the breathing zone of the potentially exposed workers were assessed by personal air monitoring. Strong correlations were found between S-PMA and tt-MA concentrations in end-of-shift samples and between either of these parameters and airborne benzene concentrations. Exposure to 1 ppm benzene (8-hr TWA) leads to an average concentration in end-of-shift samples of 21 mol S-PMA and 1.5 mmol tt-MA per mol creatinine. Of an inhaled dose of benzene, on average 0.11% (range 0.05-0.26%) was excreted as S-PMA with an apparent elimination half-life of 9.1 (standard error [SE] 0.7) hr and 3.9% (range 1.9-7.3%) as tt-MA with a half-life of 5.0 (SE 0.5) hr. Due to its longer elimination half-life, S-PMA proved a more reliable biomarker than tt-MA for benzene exposures during 12-hr shifts. Specificity of S-PMA, but not tt-MA, was sufficient to discriminate between the 14 moderate smokers and the 38 nonsmokers from the control group. The mean urinary S-PMA was 1.71 (SE 0.27) in smokers and 0.94 (SE 0.15) mol/mol creatinine in nonsmokers (p = 0.013). The mean urinary tt-MA was 0.046 (SE 0.010) in smokers and 0.029 (SE 0.013) mmol/mol creatinine in nonsmokers (p = 0.436). The inferior specificity of tt-MA was due to relatively high background values of up to 0.56 mmol/mol creatinine, which may be found in nonexposed individuals and limits the use of tt-MA to concentrations of benzene over 1 ppm (8-hr TWA). We conclude that S-PMA is superior to tt-MA as a biomarker for low-level benzene exposures because it is more specific, enabling reliable determination of benzene exposures down to 0.3 ppm (8-hr TWA), and because its longer half-life makes it more suited for biological monitoring of operators working in shifts longer than 8 hr.