In normal subjects, exercise widens the alveolar-arterial PO2 difference (P[A-a]O2) despite a more uniform topographic distribution of ventilation-perfusion (VA/Q) ratios. While part of the increase in P(A-a)O2 (especially during heavy exercise) is due to diffusion limitation, a considerable amount is caused by an increase in VA/Q mismatch as detected by the multiple inert gas elimination technique. Why this occurs is unknown, but circumstantial evidence suggests it may be related to interstitial pulmonary edema rather than to factors dependent on ventilation, airway gas mixing, airway muscle tone, or pulmonary vascular tone. In patients with lung disease, the gas exchange consequences of exercise are variable. Thus, arterial PO2 may increase, remain the same, or fall. In general, patients with advanced chronic obstructive pulmonary disease (COPD) or interstitial fibrosis who exercise show a fall in PO2. This is usually not due to worsening VA/Q relationships but mostly to the well-known fall in mixed venous PO2, which itself results from a relatively smaller increase in cardiac output than VO2. However, in interstitial fibrosis (but not COPD), there is good evidence that a part of the fall in PO2 on exercise is caused by alveolar-capillary diffusion limitation of O2 transport; in COPD (but not interstitial fibrosis), a frequent additional contributing factor to the hypoxemia of exercise is an inadequate ventilatory response, such that minute ventilation does not rise as much as does CO2 production or O2 uptake, causing arterial PCO2 to increase and PO2 to fall.