Ninety-one infants with respiratory failure secondary to primary pulmonary disease and with a birth weight of 1000 g. or over have been managed in a negative-pressure respirator (Air-Shields) over a three-year period. Of these the failure in 87 was due to respiratory distress syndrome (RDS) and in four it resulted from massive meconium aspiration. Respiratory failure was indicated initially by arterial blood gas tensions (while breathing 100% O(2)) of Po(2) <40 mm. Hg, pH <7.10 and Pco(2) >75 mm. Hg in the initial 47 cases; these levels were subsequently raised to Po(2) < 50 mm. Hg, pH <7.20 and Pco(2) >70 mm. Hg for the remainder. Fifty-four (59.3%) of the infants survived the use of the respirator and 47 of these (51.6%) were subsequently discharged alive and well. Mean time in hours to normalization of blood gas values while on the respirator were as follows: for Po(2), 10.5; for pH, 11.6; and for Pco(2), 22.6. These values indicate that the respirator is more efficient in promoting oxygenation (raising Po(2)) than ventilation (lowering Pco(2)). They also suggest that the observed acidosis is in large part secondary to the hypoxia rather than the result of co(2) retention. For the survivors the average time of total respirator dependency before commencement of weaning was 53.7 hours. All the infants were managed without the use of endotracheal tubes although the use of the respirator and/or administration of 100% oxygen were either continuous or intermittent for periods of up to two weeks. There have been no instances of so-called respirator lung disease in the survivors or in those who died, which suggests that the use of high oxygen concentration by itself is not the major factor in the pathogenesis of this complication.