An assay for the separation and detection of lipid hydroperoxides and hydrogen peroxide in biological samples using HPLC and isoluminol chemiluminescence was recently described (Y. Yamamoto, M. H. Brodsky, J. C. Baker, and B. N. Ames (1987) Anal. Biochem. 160, 7-13; Y. Yamamoto and B. N. Ames (1987) Free Rad. Biol. Med. 3, 359-361). In this paper the application of this assay to the analysis of human blood plasma is described in detail, and three compounds producing chemiluminescence that were observed in the initial studies in plasma extracted with methanol and hexane are further characterized. It is shown that various lipid hydroperoxides added to plasma are detected by the assay. In contrast, hydrogen peroxide added to plasma is rapidly degraded by endogenous catalase. Hydrogen peroxide and a second, minor compound producing chemiluminescence, which appear in the assay of the methanol and the hexane extract of plasma, respectively, appear to be generated during analysis and are not likely to be present in plasma. The third compound yielding a chemiluminescence peak, which is extracted into the hexane phase of plasma and was earlier assigned to cholesterol ester hydroperoxide, is shown to be neither a cholesterol ester nor a hydroperoxide, but the hydroquinone ubiquinol-10. As the chemiluminescence response of hydroperoxides, but not of hydroquinones, is eliminated by reducing reagents such as sodium borohydride or triphenylphosphine, such reduction should be used to confirm that any chemiluminescence producing lipid observed in the assay is a hydroperoxide, not a hydroquinone. We conclude that isolated human plasma from healthy subjects is very unlikely to contain hydrogen peroxide in concentrations greater than about 0.25 microM and does not contain lipid hydroperoxides in concentrations greater than 0.03 microM. The method described, when used with appropriate precautions, is a convenient and very sensitive assay for lipid hydroperoxides in biological tissues.