The effect of oral caffeine on resting ventilation (VE), ventilatory responsiveness to progressive hyperoxic hypercapnia (HCVR), isocapnic hypoxia (HVR), and moderate exercise (EVR) below the anaerobic threshold (AT) was examined in seven healthy adults. Ventilatory responses were measured under three conditions: control (C) and after ingestion of either 650 mg caffeine (CF) or placebo (P) in a double-blind randomized manner. None of the physiological variables of interest differed significantly for C and P conditions (P greater than 0.05). Caffeine levels during HCVR, HVR, and EVR were 69.5 +/- 11.8, 67.8 +/- 10.8, and 67.8 +/- 10.9 (SD) mumol/l, respectively (P greater than 0.05). Metabolic rate at rest and during exercise was significantly elevated during CF compared with P. An increase in VE from 7.4 +/- 2.5 (P) to 10.5 +/- 2.1 l/min (CF) (P less than 0.05) was associated with a decrease in end-tidal PCO2 from 39.1 +/- 2.7 (P) to 35.1 +/- 1.3 Torr (CF) (P less than 0.05). Caffeine increased the HCVR, HVR, and EVR slopes (mean increase: 28 +/- 8, 135 +/- 28, 14 +/- 5%, respectively) compared with P; P less than 0.05 for each response. Increases in resting ventilation, HCVR, and HVR slopes were associated with increases in tidal volume (VT), whereas the increase in EVR slope was accompanied by increases in both VT and respiratory frequency. Our results indicate that caffeine increases VE and chemosensitivity to CO2 inhalation, hypoxia, and CO2 production during exercise below the AT.