Using terephthalic acid as a water-soluble trap, we characterized hydroxyl radicals (HO?) formation by liver microsomal enzymes from isoniazid-treated rats. We found that HO? formation was entirely dependent on intact microsomal enzymes, the presence of NADPH, and iron complexed with EDTA. In contrast to the other radical traps, we found no evidence that terephthalate is a substrate for cytochrome P450. Cumene hydroperoxide, an artificial supporter of cytochrome P450-catalyzed oxidation, failed to maintain HO(.-) formation. HO(.-) formation in liver microsomes was inhibited by the HO(.-) radical scavengers: dimethyl sulfoxide (DMSO), mannitol, and citrulline. It was abolished by catalase, but not superoxide dismutase (SOD), indicating that hydrogen peroxide was the sole precursor of the HO(.-). Therefore, the generation of hydroxyl radicals by microsomal enzymes appears to be dependent on two processes: (1) the rate of hydrogen peroxide production; and (2) the availability of iron ions or other transition metals for Fenton type reactions.