Data from a pilot study on unmetabolized benzene and trans,trans muconic acid (t,t-MA) excretion in filling station attendants and unexposed controls were used to afford methodological issues in the biomonitoring of low benzene exposures (around 0.1 ppm). Urinary concentrations of benzene and t,t-MA were measured by dynamic head-space capillary GC/FID and HPLC, respectively. The accuracy of the HPLC determination of t,t-MA was assessed in terms of inter- and intra-method reliability. The adequacy of urinary t,t-MA and benzene as biological markers of low benzene exposure was evaluated by analysing the relationship between personal exposure to benzene and biomarker excretion. Filling station attendants excreted significantly higher amounts of benzene, but not of t,t-MA, than controls. Adjusting for occupational benzene exposure, smokers excreted significantly higher amounts of t,t-MA, but not of unmetabolized benzene, than nonsmokers. A comparative analysis of the present and previously published biomonitoring surveys showed a good inter-study agreement regarding the amount of t,t-MA and unmetabolized benzene excreted (about 0.1-0.2 mg/l and 1-2 micrograms/l, respectively) per unit of exposure (0.1 ppm). For each biomarker, based on the distribution of parameters observed in the pilot study, we calculated the minimum sample size required to estimate the population mean with given confidence and precision.