Diaphragmatic fatigue was induced in six normal young men inspiring against a variable alinear resistance. Breathing pattern was rigidly controlled (tidal volume 0.75 liter, 12 breaths . min-1). Fatigue was defined as an inability to continue to generate a target transdiaphragmatic pressure (Pdi = 0.65 - 0.84 Pdimax). Diaphragmatic electromyogram (EMG, esophageal electrode) and perceived effort (PE, open-ended scale) were recorded. Subjects were tested on an identical resistance inspiring air or 100% O2 in random order on different days. They were unaware of the gas mixture inspired. Mean endurance time (tlim) +/- SE for air was 4.1 +/- 1.4 min and for O2 was 8.6 +/- 2.7 min (P less than 0.005). The increased tlim in O2 was associated with a delay in onset of EMG changes heralding diaphragmatic fatigue and a decrease in PE at any time during the study compared with the level of PE in air. Arterial O2 saturation (ear oximeter) remained at the resting level of 99.0 +/- 0.2% in O2 and decreased from the resting level of 97.2 +/- 0.2% by 2.8 +/- 0.7% (P less than 0.01) in air. The end-tidal CO2 fraction increased to a similar degree in air and O2 studies. We conclude that when breathing pattern, minute ventilation, and Pdi are held constant during inspiratory resistive loading, breathing O2 delays the onset of diaphragm fatigue and decreases PE.