One century after the introduction of the oxygen flow meter into clinical practice, we have developed a device, FreeO(2), that automatically titrates the oxygen flow delivered to spontaneously breathing patients, with the aim of maintaining a stable S(pO(2)). We evaluated this system in healthy subjects during induced hypoxemia. Hypoxemia was induced in 10 healthy subjects while breathing a gas mixture of variable F(IO(2)) (air + nitrogen). Each subject performed 3 hypoxemic challenges with the addition, in a random order, of either: air with constant flow (1.5 L/min); oxygen with constant flow (1.5 L/min); or automatic oxygen flow titration. Subjects were blinded to the intervention. Oxygen flow, S(pO(2)), end-tidal CO(2), respiratory rate, and heart rate were recorded every second. The primary outcome was the time with S(pO(2)) between 92% and 96%. The S(pO(2)) target (92-96%) was achieved a median of 26.0%, 36.8%, and 66.5% (P < .001) of the time with air, constant oxygen, and automated oxygen titration, respectively. Severe oxygen desaturations (S(pO(2)) < 88%) were respectively observed at a median of 33.7%, 12.7%, and 0.4% of the time (P < .001). Hyperoxia was present a median of 4.1%, 39.1%, and 14.5% of the time (P < .001). Tachycardia was present with air and with constant oxygen flow, but not while using automated oxygen titration. These results were obtained with a mean and maximal oxygen flow of 1.3 L/min and 7.6 L/min with the automated titration. In this model of induced hypoxemia, the FreeO(2) system that automatically titrates the oxygen flow was more efficient at maintaining the S(pO(2)) target, while ensuring a statistically significant reduction in the rates of severe hypoxemia and hyperoxia, in comparison with air or constant oxygen flow. These beneficial results were obtained with less oxygen, in comparison to a constant oxygen flow.