During exercise by healthy mammals, alveolar ventilation and alveolar-capillary diffusion increase in proportion to the increase in metabolic rate to prevent PaCO2 from increasing and PaO2 from decreasing. There is no known mechanism capable of directly sensing the rate of gas exchange in the muscles or the lungs; thus, for over a century there has been intense interest in elucidating how respiratory neurons adjust their output to variables which can not be directly monitored. Several hypotheses have been tested and supportive data were obtained, but for each hypothesis, there are contradictory data or reasons to question the validity of each hypothesis. Herein, we report a critique of the major hypotheses which has led to the following conclusions. First, a single stimulus or combination of stimuli that convincingly and entirely explains the hyperpnea has not been identified. Second, the coupling of the hyperpnea to metabolic rate is not causal but is due to of these variables each resulting from a common factor which link the circulatory and ventilatory responses to exercise. Third, stimuli postulated to act at pulmonary or cardiac receptors or carotid and intracranial chemoreceptors are not primary mediators of the hyperpnea. Fourth, stimuli originating in exercising limbs and conveyed to the brain by spinal afferents contribute to the exercise hyperpnea. Fifth, the hyperventilation during heavy exercise is not primarily due to lactacidosis stimulation of carotid chemoreceptors. Finally, since volitional exercise requires activation of the CNS, neural feed-forward (central command) mediation of the exercise hyperpnea seems intuitive and is supported by data from several studies. However, there is no compelling evidence to accept this concept as an indisputable fact.