The thiobarbituric acid (TBA) reactivity of human plasma was studied to evaluate its adequacy in quantifying lipid peroxidation as an index of systemic oxidative stress. Two spectrophotometric TBA tests based on the use of either phosphoric acid (pH 2.0, method A) or trichloroacetic plus hydrochloric acid (pH 0.9, method B) were employed with and without sodium sulfate (SS) to inhibit sialic acid (SA) reactivity with TBA. To correct for background absorption, the absorbance values at 572 nm were subtracted from those at 532 nm, which represent the absorption maximum of the TBA:MDA adduct. Method B gave values of TBA-reactive substances (TBARS) 2-fold higher than those detected with method A. SS lowered TBARS by about 50% with both methods, indicating a significant involvement of SA in plasma TBA reactivity. Standard SA, at a physiologically relevant concentration of 1.5 mM, reacted with TBA, creating interference problems, which were substantially eliminated by SS plus correction for background absorbance. When method B was carried out in the lipid and protein fraction of plasma, SS inhibited by 65% TBARS formation only in the latter. Protein TBARS may be largely ascribed to SA-containing glycoproteins and, to a minor extent, protein-bound MDA. Indeed, EDTA did not affect protein TBARS assessed in the presence of SS. TBA reactivity of whole plasma and of its lipid fraction was instead inhibited by EDTA, suggesting that lipoperoxides (and possibly monofunctional lipoperoxidation aldehydes) are involved as MDA precursors in the TBA test. Pretreatment of plasma with KI, a specific reductant of hydroperoxides, decreased TBARS by about 27%. Moreover, aspirin administration to humans to inhibit prostaglandin endoperoxide generation reduced plasma TBARS by 40%. In conclusion, reaction conditions affect the relationship between TBA reactivity and lipid peroxidation in human plasma. After correction for the interfering effects of SA in the TBA test, 40% of plasma TBARS appears related to in vivo generated prostaglandin endoperoxides and only about 60% to lipoperoxidation products. Thus, the TBA test is not totally specific to oxidant-driven lipid peroxidation in human plasma.