1. Transient changes of heart rate (HR) and ventilation were recorded following step changes in alveolar gas composition in three healthy subjects. From a steady state of normo- or slightly hypercapnic hypoxia (PA,CO2 38-46 torr, PA,O2 50-60 torr) arterial chemoreceptor stimulation was transiently relieved by breathing a CO2-free mixture for two breaths, either pur O2 (causing a fall in PA,CO2 and a rise in PA,O2; O2 test) or a low O2 mixture (causing a fall in PA,CO2 without any change in PA, O2; CO2 test). For both test types ventilation was either allowed to change freely ('free-breathing' tests) or was consciously maintained at the pre-test level by the subjects ('controlled-breathing tests). The circulatory delay from the lungs to the ear was measured with a sensitive ear oximeter. 2. In all 'free-breathing' tests ventilation decreased significantly after a mean latency of 5.2 sec; the average lung-ear circulation time was 4.9 sec. HR increased slightly above pre-test levels in eighty-one of one hundred and four tests of all types, the changes being significant after a latency identical to that of the ventilatory changes. Except in the 'controlled-breathing' CO2 tests this early tachycardia was followed by a decrease in HR within the following 5-6 sec. 3. These findings indicate that the primary effect of withdrawal of arterial chemoreceptor stimulation in conscious man as in the anesthetized animal is tachycardia. The secondary development of bradycardia in 'free-breathing' CO2 tests is probably due to the operation of a lung reflex sensing changes in ventilation. The absence of bradycardia in 'controlled-breathing' CO2 tests and its presence in 'controlled-breathing' O2 tests, finally, suggest that relief of systemic hypoxia causes a slowing of the heart not due to lung reflexes but to some other mechanism which operates with a latency nearly twice as long as the arterial chemoreflex.