The effects of isocapnic hyperventilation (A) and normoventilation (B) on PaCO2, PaO2 and A-aDO2 were compared in 102 patients undergoing elective surgery, randomized into two comparable groups A and B. Cases for thoracic, high abdominal and intracranial surgery were excluded, as well as patients with clinically evident pulmonary pathology. A volumetric ventilator was used in association with three different breathing systems (A: Bain system and circle system without CO2 absorption; B: circle system with CO2 absorption). The groups were comparable, except for percentage of overweight: 75% in group A and 56% in group B. Overweight was defined as weight above the mean ideal weight of 100%, and obesity as a weight above 120% the mean ideal weight. Blood gases were sampled 1) preoperatively, 2) 15 min and 3) 60 min after the beginning of mechanical ventilation, 4) postoperatively, 90 min after extubation, without supplemental oxygen. The preoperative mean PaO2 values were 79 +/- 11.4 mmHg (A) and 82.5 +/- 13.2 mmHg (B); the PaCO2 were 37.2 +/- 3.7 mmHg (A) and 37.2 +/- 3.8 mmHg (B). During surgery, PaO2 was distinctly higher (p less than 0.01) in group A than in group B (on average 15-20 mmHg higher), indicating the favourable effect of great tidal volumes on gas exchange. Correspondingly, the A-aDO2 was less increased in group A than in group B (p less than 0.01). At 15 min, 33% of the patients were hypocapnic (PaCO2 less than 35 mmHg): one case in group A and three in group B could be classed as severe hypocapnia (PaCO2 = 25-30 mmHg).(ABSTRACT TRUNCATED AT 250 WORDS)