Adenosine infusion (100 micrograms X kg-1 X min-1) in humans stimulates ventilation but also causes abdominal and chest discomfort. To exclude the effects of symptoms and to differentiate between a central and peripheral site of action, we measured the effect of adenosine infused at a level (70-80 micrograms X kg-1 X min-1) below the threshold for symptoms. Resting ventilation (VE) and progressive ventilatory responses to isocapnic hypoxia and hyperoxic hypercapnia were measured in six normal men. Compared with a control saline infusion given single blind on the same day, adenosine stimulated VE [mean increase: 1.3 +/- 0.8 (SD) l/min; P less than 0.02], lowered resting end-tidal PCO2 (PETCO2) (mean fall: -3.9 +/- 0.9 Torr), and increased heart rate (mean increase: 16.1 +/- 8.1 beats/min) without changing systemic blood pressure. Adenosine increased the hypoxic ventilatory response (control: -0.68 +/- 0.4 l X min-1 X %SaO2-1, where %SaO2 is percent of arterial O2 saturation; adenosine: -2.40 +/- 1.2 l X min-1 X %SaO2-1; P less than 0.01) measured at a mean PETCO2 of 38.3 +/- 0.6 Torr but did not alter the hypercapnic response. This differential effect suggests that adenosine may stimulate ventilation by a peripheral rather than a central action and therefore may be involved in the mechanism of peripheral chemoreception.