Gases are a class biologically active species that are normally transported to the blood by the lungs. Examples include oxygen and other metabolites, as well as nitrous oxide and other anesthetics. But what happens when the lungs are not functioning adequately? Respiratory distress syndrome and other forms of lung injury can result in acute hypoxia, which can rapidly lead to severe morbidity and mortality. Extracorporeal membrane oxygenation (ECMO) currently is the standard clinical treatment for acute hypoxia, but the technique is cumbersome and problematic for certain patient populations, including infants. A means of directly injecting oxygen into the bloodstream, thus precluding the need for an extracorporeal loop, is highly desired in the critical care setting. Such a technology could augment or even replace ECMO for some cases. The approach we have taken is to use oxygen-filled microbubbles for direct intravenous oxygenation. Microbubbles are 0.1 to 10 micron diameter gas spheres, coated and stabilized by a lipid monolayer shell, and surrounded by an aqueous medium. Results will be presented which show that oxygen microbubbles can be reproducibly generated and concentrated to 70 vol%. A dose-response relationship was measured for microbubble suspensions combined in vitro with venous blood. The oxygen carrying and release capability is strongly dependent on lipid composition. Lipids above their main phase transition temperatures do not encapsulate oxygen for meaningful time scales. Those in the gel phase encapsulate and release oxygen efficiently, and an optimum composition was found. Phenomena governing microbubble dissolution in blood will be discussed, including modeling of gas exchange kinetics and excess lipid collapse and shedding. Preliminary results are promising and suggest improvements for future testing in vivo.