We examined the role of metabolic CO2 production in the hyperpnea of muscular exercise by comparing the response of alveolar ventilation to moderate levels of exercise with the response to venous infusion of CO2 at rest. Studies were performed in four awake sheep that were trained to run on a treadmill. The sheep had been cannulated for veno-venous extracorporeal perfusion so that CO2 could be infused into the peripheral venous blood through membrane lungs in the perfusion circuit. The sheep breathed room air through an endo-tracheal tube inserted through a tracheostomy, and samples of expired gas were collected for measurement of the rates of CO2 production and O2 consumption. All measurements were made in the steady state. In each of the four sheep, the relationship between alveolar ventilation and the rate of CO2 production could be described by a single linear function (r greater than 0.99; P less than 0.001), regardless of whether CO2 production was increased by exercise, venous CO2 infusion, or combinations of both procedures. This relationship applied for values of CO2 production up to 350% of control. In contrast, no unique relationship was found between the rate of alveolar ventilation and either the rate of O2 consumption, cardiac output, or mixed venous blood gas pressures. The findings indicate that the hyperpnea of mild to moderate steady-state exercise can be attributed to the associated increase in the rate of CO2 production. Therefore, there is no need to invoke obligatory nonmetabolic stimuli to account for the ventilatory response to steady-state exercise.