In aquaculture, oxygen transfer and carbon dioxide stripping are the first limiting factors to fish rearing intensification. In this study we measured the O-2 and CO2 mass transfer coefficient (K(L)a) for a vacuum airlift in fresh (<1% salinity) and sea water (35 parts per thousand salinity) recirculating aquaculture systems (RAS). The airlift was composed of two concentric tubes: an inner riser tube and an external downcomer tube and can be adjusted at three different heights: 2,4 or 6 m. Several types of air injectors were tested, delivering different sizes of bubble swarms depending on their porosity and functioning conditions (low or high injection pressure), with air flows varying from 0 to 80 L min(-1). Experiments were also carried out at different water circulation velocities and with cold (7 degrees C) and warm water (22 degrees C). The best transfer coefficient (K(L)a) value was obtained at a high air flow rate, a high temperature and with reduced bubble size. Results showed that KLa was not affected by water salinity, but it was slightly affected by water flow (Q), airlift inner pipe length and vacuum. The presence of vacuum reduces gas solubility in water and facilitates CO2 stripping. The comparison between O-2 and CO2 transfers showed that higher KLa values were obtained for O-2 than for CO2 in fresh and sea water, probably due to chemical reactions between the CO2 and water. For RAS, the vacuum airlift provides a Standard Aeration Efficiency (SAE) of 1.13 kgO(2) kW h(-1) and a Standard Stripping Efficiency (SSE) of 1.8 kgO(2) kW h(-1) or 0.023 kgCO(2) kW h(-1). In rearing water, CO2 and O-2 transfers were negatively affected when feed was added. An empirical model for CO2 mass transfer coefficient prediction was developed and calibrated. Simulation shows a good correlation between predicted and measured values (R-2 = 0.87). (C) 2011 Elsevier B.V. All rights reserved.