The correct tidal volume during cardiopulmonary resuscitation (CPR) is presently debated. While the European Resuscitation Council (ERC) and American Heart Association (AHA) previously recommended a tidal volume of 800-1200 ml, the ERC has recently reduced this to 400-600 ml. In a prospective, randomised study of 17 non-traumatic out-of-hospital cardiac arrest patients intubated and mechanically ventilated 12 min(-1) with 100% oxygen, we have therefore compared arterial blood gases generated with tidal volumes of 500 and 1000 ml. Mean time from cardiac arrest to arrival of the ambulance was 13+/-8 and 14+/-8 min in the two groups, respectively. Arterial blood samples were taken percutaneously 5 and 10-15 min after onset of the mechanical ventilation and analysed instantly. Pa(CO(2)) was significantly higher for a tidal volume of 500 than 1000 ml at both 5 and 10-15 min, 7.48+/-2.23 versus 3.70+/-0.83 kPa (P=0.002) and 7. 45+/-1.19 versus 3.98+/-1.58 kPa (P<0.001). The pH was lower for 500 than 1000 ml at 10-15 min, 7.01+/-0.10 versus 7.20+/-0.17 (P=0.034), with a strong trend in the same direction at 5 min (P=0.06). There was adequate oxygenation with no differences in Pa(O(2)) or BE at any time between the two groups, and no significant differences in any blood gas variables between the 5- and 10-15-min samples. We conclude that arterial normocapnia is not achieved with either tidal volume during advanced life support with non-rebreathing ventilation at 12 min(-1). What ventilation volume is required for CO(2) removal and oxygenation during basic life support with mouth-to-mouth ventilation cannot be extrapolated from the present data. In that situation the risk of gastric inflation, regurgitation and aspiration must also be taken into account.