To assess arteriovenous differences in acid-base status, we measured the pH and partial pressure of carbon dioxide (PCO2) in blood drawn simultaneously from the arterial and central venous circulations in 26 patients with normal cardiac output, 36 patients with moderate and 5 patients with severe circulatory failure, and 38 patients with cardiac or cardiorespiratory arrest. The patients with normal cardiac output had the expected arteriovenous differences: venous pH was lower by 0.03 unit, and venous PCO2 was higher by 0.8 kPa (5.7 mm Hg). These differences widened only slightly in those with moderate cardiac failure. Additional simultaneous determinations in mixed venous blood from pulmonary arterial catheters were nearly identical to those in central venous blood. In the five hypotensive patients with severe circulatory failure there were substantial differences between the mean arterial and central venous pH (7.31 vs. 7.21) and PCO2 (5.8 vs. 9.0 kPa [44 vs. 68 mm Hg]). Large arteriovenous differences were present during cardiac arrest in patients whose ventilation was mechanically sustained, whether sodium bicarbonate had been administered (pH, 7.27 vs. 7.07; PCO2, 5.8 vs. 8.6 kPa [44 vs. 65 mm Hg]) or not (pH, 7.36 vs. 7.01; PCO2, 3.7 vs. 10.2 kPa [28 vs. 76 mm Hg]). By contrast, in patients with cardiorespiratory arrest, large arteriovenous differences were noted only when sodium bicarbonate had been given (pH, 7.24 vs. 7.01; PCO2, 9.5 vs. 16.9 kPa [71 vs. 127 mm Hg]). We conclude that both arterial and central venous blood samples are needed to assess acid-base status in patients with critical hemodynamic compromise. Although information about arterial blood gases is needed to assess pulmonary gas exchange, in the presence of severe hypoperfusion, the hypercapnia and acidemia at the level of the tissues are detected better in central venous blood.