The thiobarbituric acid (TBA) reaction with malondialdehyde (MDA) is commonly used to measure free radical- mediated oxidative changes in lipids containing polyunsaturated fatty acids (PUFA). This test has had a broad range of applications since it was introduced in 1949; among others, researchers have used this test to evaluate toxic outcomes of chemical exposure in many environmental and occupational settings. However, the test has been criticized because of its low specificity. It has been reported that several endogenous or exogenous compounds, including drugs and metals, may interfere with a photometric or fluorimetric assay. The wide use of this test in occupational toxicology to measure lipid peroxidation (LPO) in humans led us to investigate whether simultaneous medications with drugs or exposure to metals interfere with the assay. We show here that, in the presence of bismuth or medazepam, the test gives erroneous results in a spectrophotometric assay. Among the tested 45 compounds, bismuth inhibits the formation of the MDA-TBA complex. In contrast, medazepam caused 2-fold higher absorbance compared to the absorbance obtained in the absence of the drug. UV spectrum analysis revealed that another unknown product is formed with an absorption maximum of 458 nm. In addition, medazepam yielded a second product with TBA whose absorption spectrum was essentially identical to that of the MDA standard. We have succeeded in separating this product and quantifying the MDA-TBA complex accurately by a HPLC method. Though it requires more sophisticated equipment, we suggest using HPLC for MDA quantification in biological samples, which might contain potentially interfering compounds.