Single and double hollow fiber intravascular gas exchangers were evaluated in an extracorporeal veno-venous bypass circuit (right atrium to pulmonary artery) including a tubular blood chamber (mimicking caval veins with an inner diameter of 26 mm) for evaluation of the membrane surface area/host vessel diameter gas transfer relationships. Six bovine experiments (body wt: 68 +/- 4 kg) with staged ex vivo blood flows of 1, 2, 3, and 4 L/min and a device oxygen inflow of 0, 3, and 6 L/min (0 or 3 L/min/device) were performed. Total oxygen transfer at a blood flow of 1 L/min was 33 +/- 4 ml/ min for a gas flow of 3 L/min (one device) vs 60 +/- 25 ml/ min for a gas flow of 6 L/min (two devices); at a blood flow of 2 L/min, the corresponding oxygen transfer was 46 +/- 16 ml/min for a gas flow of 3 L/min vs 95 +/- 44 ml/min for a gas flow of 6 L/min; at a blood flow of 3 L/min, the corresponding oxygen transfer was 48 +/- 24 ml/min for a gas flow of 3 L/ min vs 92 +/- 37 ml/min for a gas flow of 6 L/min (p < 0.01 for comparison of areas under the curves). Total carbon dioxide transfer at a blood flow of 1 L/min was 47 +/- 18 ml/min for a gas flow of 3 L/min vs 104 +/- 26 ml/min for a gas flow of 6 L/min; at a blood flow of 2 L/min, the corresponding carbon dioxide transfer was 59 +/- 19 ml/min for a gas flow of 3 L/ min vs 129 +/- 39 ml/min for a gas flow of 6 L/min; at a blood flow of 3 L/min, the corresponding carbon dioxide transfer was 60 +/- 22 ml/min for a gas flow of 3 L/min vs 116 +/- 49 ml/min for a gas flow of 6 L/min (p < 0.01). For the given setup, the blood flow/gas transfer relationship is non linear, and a plateau is achieved at a blood flow of 2.5 L/min for O2 and CO2. Doubling membrane surface area and consecutively sweeping gas flows result in doubling of gas transfers at all tested blood flows. However, increased membrane surface area and blood flow produce a higher pressure drop that in turn limits the fiber density that can be used clinically.