Instrumental dead space wash-out can be used to improve carbon dioxide clearance. The aim of this study was to define, using a bench test, an optimal protocol for long-term use, and to assess the efficacy of this technique in neonates.A bench test with an artificial lung model, and an observational prospective study. Dead space wash-out was performed by continuous tracheal gas insufflation (CTGI), via six capillaries molded in the wall of a specially designed endotracheal tube, in 30 preterm neonates with hyaline membrane disease.Neonatal intensive care unit of a regional hospital.The bench test study showed that a CTGI flow of 0.5 l/ min had the optimal efficacy-to-side-effect ratio, resulting in a maximal or submaximal efficacy (93 to 100%) without a marked increase in tracheal and CTGI circuit pressures. In the 30 newborns, 15 min of CTGI induced a significant fall in arterial carbon dioxide tension (PaCO2), from 45 +/- 7 to 35 +/- 5 mmHg (p = 0.0001), and in 14 patients allowed a reduction in the gradient between Peack inspirating pressure and positive end-expiratory pressure from 20.8 +/- 4.6 to 14.4 +/- 3.7 cmH2O (p < 0.0001) while keeping the transcutaneous partial pressure of carbon dioxide constant. As predicted by the bench test, the decrease in PaCO2 induced by CTGI correlated well with PaCO2 values before CTGI (r = 0.58, p < 0.002) and with instrumental dead space-to-tidal volume ratio (r = 0.54, p < 0.005).CTGI may be a useful adjunct to conventional ventilation in preterm neonates with respiratory disease, enabling an increase in CO2 clearance or a reduction in ventilatory pressure.