1. Venous admixture/cardiac output ratio (Qva/Qt) has been measured in twenty-four healthy volunteer subjects of both sexes aged 20-71 years, at rest and during the steady state of treadmill exercise at two rates of work, and breathing air and breathing oxygen. 2. With oxygen breathing, Qva/Qt was considerably less during exercise than during the time subjects were taking either normal or deep breaths of oxygen at rest, and did not significantly increase with the intensity of exercise. It is postulated that the increase in ventilation during exercise opens most or all of those alveoli which, during oxygen breathing at rest, close because of critically low ventilation/perfusion (V/Q) ratios. 3. With air breathing, Qva/Qt fell from rest to exercise (especially in older subjects), presumably due to improved ventilation of alveoli at the lung bases. With an increase in work rate Qva/Qt increased in all age groups. This increase was not due to increase in the shunt fraction (Qs/Qt), nor to limitation of diffusing capacity; it arose from an increase in V/Q variance. 4. Equations have been derived for the prediction of normal Qva/Qt during exercise, with or without correction for the effects of increasing pulmonary capillary temperature. These effects do not materially influence the accuracy of prediction, but may be relevant to some of the interpretations. In particular, they provide a further indication that Qs/Qt probably cannot be measured by breathing oxygen at rest, even in deep breathing.