In resting conscious dogs physiological dead space was calculated using the Bohr equation and measurements of arterial and mixed expired carbon dioxide tension. Whenever dogs inhaled carbon dioxide mixtures (5-10%) that had normal or low oxygen concentrations, the calculated dead space became negative. This paradox was based on the fact that the mixed expired carbon dioxide tension in resting hypercapnic dogs. Under these circumstances carbon dioxide was produced from the lung as measured by gas analyses and blood analyses. By the lung as measured by gas analyses and blood analyses. By reasoning this implies that "alveolar" carbon dioxide tension was higher than pulmonary venous carbon dioxide tension. The negative carbon dioxide gradient persisted at 14 days of chronic hypercapnia and reverted to normal within 10 min of breathing air after chronic hypercapnia. These findings suggest that the exchange of carbon dioxide in the lung cannot be explained solely on the basis of passive diffusion.