Free-radical production during the metabolism of various xenobiotics represents a frequent mechanistic explanation for their toxicity. We tested the hypothesis of production of free radicals from two solvents, diethylether and trichloroethylene (TRI), and from two metabolites of TRI, namely trichloroethanol (TCE) and trichloroacetic acid (TCA). The formation of free radicals was detected by electron spin resonance spectroscopy (ESR), using a spin-trapping agent, alpha-(4-pyridyl-1-oxide)-N-tert-butyl-nitrone (POBN). Two experimental models were used. The first was a chemical model using Fenton's reagent, a mixture of Fe(II)-chelator and H2O2, for which the normal reaction is OH. production, and the second, a preparation from rat liver and brain microsomes containing NADPH and achieving enzymatic oxidation of the solvents. After addition of diethylether, free-radical production was demonstrated under the two experimental conditions. This free radical probably derived from the parent molecule by hydrogen abstraction. TRI and TCE additions to the Fenton system suppressed normal OH. production whereas this production was increased after TCA addition. The addition of TCE to the microsomal preparations was followed by free-radical production which could derive either from the parent molecule or from other sources, e.g. from membrane degradation, with a preference for the first hypothesis because of the characteristics of the signal. This result was not observed after addition of TRI or TCA. In conclusion, these preliminary results confirm the validity of the hypothesis of production of free radicals from diethylether, but they are less consistent for TRI as this production was observed only after addition of TCE; this result is interesting, however, as TCE is considered to play a major role in the toxicity observed after TRI exposure in humans.