1. Some glycidyl ethers (GE) have been shown to be direct mutagens in short-term in vitro tests and consequently GE are considered to be potentially mutagenic in vivo. However, GE may be metabolically inactivated in the body by two different enzymatic routes: conjugation of the epoxide moiety with the endogenous tripeptide glutathione (GSH) catalysed by glutathione S-transferase (GST) or hydrolysis of the epoxide moiety catalysed by epoxide hydrolase (EH). 2. The metabolic inactivation of five different GE, the diglycidyl ethers of bisphenol A (BADGE), 4,4'-dihydroxy-3,3',5,5'-tetramethylbiphenyl (Epikote YX4000) and 1,6-hexanediol (HDDGE) and the GE of 1-dodecanol (C12GE) and o-cresol (o-CGE), has been studied in subcellular fractions of human, C3H mouse and F344 rat liver and lung. 3. All GE were chemically very stable and resistant to aqueous hydrolysis, but were rapidly hydrolysed by EH in cytosolic and microsomal fractions of liver and lung. The aromatic GE were very good substrates for EH. In general, microsomal EH is more efficient than cytosolic EH in hydrolysis of GE, and human microsomes are more efficient than rodent microsomes. 4. The more water-soluble GE, o-CGE and HDDGE, were good substrates for GST whereas the more lipophilic GE, YX4000 and C12GE, were poor substrates for GST. In general, rodents are more efficient in GSH conjugation of GE than humans. 5. In general, the epoxide groups of YX4000 are the most and those of HDDGE the least efficiently inactivated of the five GE under study. For the other three GE no general trend was observed: the relative efficiency of inactivation varied with organ and species. 6. The large variation in metabolism observed with five representative GE indicate that GE have variable individual properties and should not be considered as a single, homogenous class of compounds.