Many apparently healthy individuals experience pulmonary gas exchange limitations during exercise, and the term "exercise induced arterial hypoxemia" (EIAH) has been used to describe the increase in alveolar-arterial difference for oxygen (AaDO2), which combined with a minimal alveolar hyperventilatory response, results in a reduction in arterial PO2. Despite more than two decades of research, the mechanisms of pulmonary gas exchange limitations during exercise are still debated. Using data in 166 healthy normal subjects collated from several previously published studies it can be shown that approximately 20% of the variation in PaO2 between individuals can be explained on the basis of variations in alveolar ventilation, whereas variations in AaDO2 explain approximately 80%. Using multiple inert gas data the relative contributions of ventilation-perfusion ("VA/Q") inequality and diffusion limitation to the AaDO2 can be assessed. During maximal exercise, both in individuals with minimal (AaDO2 < 20 Torr, x = 13 +/- 5, means +/- SD, n = 35) and moderate to severe (AaDO2= 25-40 Torr, x = 33 +/- 6, n = 20) gas exchange limitations, VA/Q inequality is an important contributor to the AaDO2. However, in subjects with minimal gas exchange impairment, VA/Q inequality accounts for virtually all of the AaDO2 (12 +/- 6 Torr), whereas in subjects with moderate to severe gas exchange impairment it accounts for less than 50% of the AaDO2 (15 +/- 6 Torr). Using this framework, the difficulties associated with unraveling the mechanisms of pulmonary gas exchange limitations during exercise are explored, and current data discussed.