In human plasma, 11-dehydrothromboxane (TX) B2 is a major long lived metabolite (t1/2 45 min) formed from infused TXB2, the hydration product of biologically active TXA2. Plasma concentrations of TXB2 itself are readily confounded by ex vivo platelet activation and, theoretically, an enzymatic derivative of this compound, not subject to formation in whole blood, would more accurately reflect TXA2 formation in vivo. To address this hypothesis, we developed a sensitive assay for both 11-dehydro-TXB2 and TXB2, using capillary gas chromatography/negative-ion chemical ionization mass spectrometry. We established that whole blood possesses a minor capacity to form 11-dehydro-TXB2, attributable to nonenzymatic formation in erythrocytes. However, the nonenzymatic formation of 11-dehydro-TXB2 was not a practical limitation to its use as an index of TX biosynthesis. Blood was drawn from healthy volunteers (i) via an indwelling catheter at the time of insertion and at 30, 60, 90, 180, and 240 min thereafter and (ii) via separate venipunctures at 0 time and at 90 and 240 min thereafter. Plasma TXB2 drawn via the catheter at baseline (66 +/- 63 pg/ml) was substantially greater than the maximal estimate of endogenous TXB2 (1-2 pg/ml) in plasma [Patrono, C., Ciabattoni, G., Pugliese, F., Perruci, A., Blair, I. A. & FitzGerald, G. A. (1986) J. Clin. Invest. 77, 590-594] and increased in magnitude and variance over time (339 +/- 247 pg/ml at 240 min). By contrast, 11-dehydro-TXB2 did not change significantly in the sequential catheter samples or in the samples drawn by separate venipuncture. Basal plasma concentrations in volunteers were depressed by pretreatment with 325 mg of aspirin. Furthermore, the range of concentrations in patients with severe atherosclerosis in whom urinary 2,3-dinor-TXB2 was increased was significantly higher (5-50 pg/ml, P less than 0.01) than in healthy subjects (0.9-1.8 pg/ml). Concentrations of 11-dehydro-TXB2 were increased in patients who had recently suffered a pulmonary embolism to a greater extent than either the 11-dehydro-13,14-dihydro-15-keto-TXB2 or the 2,3-dinor-TXB2 metabolites in plasma. These results indicate that plasma TXB2 is readily confounded by platelet activation ex vivo. Measurement of enzymatic metabolites of TXB2 minimizes this problem. The 11-dehydro metabolite is the most appropriate analytic target to detect phasic release of TXA2 in the human circulation, such as might occur in human syndromes of platelet activation.