Many studies have investigated the effects of glutathione S-transferase (GST) polymorphisms on cancer incidence in people exposed to carcinogenic polycyclic aromatic hydrocarbons (PAHs). The basis for this is that the carcinogenic bay region diol epoxide metabolites of several PAH are detoxified by GSTs in in vitro studies. However, there are no reports in the literature on the identification in urine of the mercapturic acid metabolites that would result from this process in humans. We addressed this by developing a method for quantitation in human urine of mercapturic acids which would be formed from angular ring diol epoxides of phenanthrene (Phe), the simplest PAH with a bay region, and a common environmental pollutant. We prepared standard mercapturic acids by reactions of syn- or anti-Phe-1,2-diol-3,4-epoxide and syn- or anti-Phe-3,4-diol-1,2-epoxide with N-acetylcysteine. Analysis of human urine conclusively demonstrated that the only detectable mercapturic acid of this type--N-acetyl-S-(r-4,t-2,3-trihydroxy-1,2,3,4-tetrahydro-c/t-1-phenanthryl)-L-cysteine (anti-PheDE-1-NAC)--was derived from the 'reverse diol epoxide', anti-Phe-3,4-diol-1,2-epoxide, and not from the bay region diol epoxides, syn- or anti-Phe-1,2-diol-3,4-epoxide. Levels of anti-PheDE-1-NAC in the urine of 36 smokers were (mean +/- SD) 728 +/- 859 fmol/ml urine. The results of this study provide the first evidence for a mercapturic acid of a PAH diol epoxide in human urine, but it was not derived from a bay region diol epoxide as molecular epidemiologic studies have presumed, but rather from a reverse diol epoxide, representative of metabolites with little if any carcinogenic activity. These results demonstrate the need for integration of genotyping and phenotyping information in molecular epidemiology studies.