The correlation between low level time-weighted average (TWA) atmospheric xylene exposure (p.p.m.) and urinary methylhippuric acid (MHA) expressed per gram of creatinine was examined. Subjects were recruited from workplaces that utilized xylene. Ambient monitoring of o-, m- and p-xylene isomers was carried out using passive diffusion vapour monitors. Adjusted (post-shift minus pre-shift) and post-shift urinary levels of xylene metabolites (2-, 3- and 4-MHA) were determined by GC-MS. Twenty subjects were recruited into the study. Total xylene TWA exposures were 3.36 +/- 3.63 p.p.m. (mean +/- SD) with a range of 0.03-14.44 p.p.m. The r(2) values for the regression equations between xylene exposure and individual and total adjusted MHA isomers were 0.390, 0.709, 0.677 and 0.631 for o-, m-, p- and total xylenes, respectively, which was greater than the respective correlations between non-adjusted samples. In conclusion, biological monitoring of occupational xylene exposure at levels <15 p.p.m. using urinary MHA showed a good correlation with atmospheric levels and is a valid complement to ambient monitoring. Even though occupational xylene exposure in the workplaces studied was generally low, MHA was found in the pre-shift urine of all workers and the use of adjusted values showed modest improvements in correlations. Recent exposure prior to sampling, either from occupational or non-occupational sources, should be considered when biological monitoring of xylene is undertaken. Extrapolation of data from this study predicted a MHA concentration in post-shift urine of 1.3 g/g creatinine after exposure to a TWA of 100 p.p.m. xylene.