cis -9,10-Epoxystearic acid was used as a tool to probe the active sites of epoxide hydrolases (EHs) of mammalian and plant origin. We have compared the stereochemical features of the hydrolysis of this substrate catalysed by soluble and membrane-bound rat liver EHs, by soluble EH (purified to apparent homogeneity) obtained from maize seedlings or celeriac roots, and by recombinant soybean EH expressed in yeast. Plant EHs were found to differ in their enantioselectivity, i.e. their ability to discriminate between the two enantiomers of 9,10-epoxystearic acid. For example, while the maize enzyme hydrated both enantiomers at the same rate, the EH from soybean exhibited very high enantioselectivity in favour of 9 R ,10 S -epoxystearic acid. This latter enzyme also exhibited a strict stereoselectivity, i.e. it hydrolysed the racemic substrate with a very high enantioconvergence, yielding a single chiral diol product, threo -9 R ,10 R -dihydroxystearic acid. Soybean EH shared these distinctive stereochemical features with the membrane-bound rat liver EH. The stereochemical outcome of these enzymes probably results from a stereoselective attack by the nucleophilic residue on the oxirane ring carbon having the ( S )-configuration, leading to the presumed (in plant EH) covalent acyl–enzyme intermediate. In sharp contrast, the reactions catalysed by cytosolic rat liver EH exhibited a complete absence of enantioselectivity and enantioconvergence; this latter effect might be ascribed to a regioselective formation of the acyl–enzyme intermediate involving C-10 of 9,10-epoxystearic acid, independent of its configuration. Thus, compared with soybean EH, the active site of rat liver soluble EH displays a very distinct means of anchoring the oxirane ring of the fatty acid epoxides, and therefore appears to be a poor model for mapping the catalytic domain of plant EHs.