A prospective study of 99 premature infants with severe respiratory distress syndrome who were randomly assigned to receive diuretic treatment with either furosemide or chlorothiazide was analyzed to examine the relationship of diuretic administration and diuresis to survival and to the duration and degree of mechanical ventilatory support. Subjects were given a diuretic, usually beginning on the second or third day of life, if they had not initiated the expected spontaneous diuresis and did not show pulmonary improvement. Infants given furosemide experienced a postnatal weight loss nearly identical to that in infants who were deemed not to need a diuretic; infants given chlorothiazide lost weight more slowly and had significantly greater body weight on postnatal days 4 and 5. Four factors were independently correlated with improved survival: furosemide usage, high birth weight, low initial mean airway pressure, and the absence of intraventricular hemorrhage. Ventilator mean airway pressure on the seventh day of life and duration of mechanical ventilation were both related to diuresis. These data provide additional evidence for the importance of water homeostasis in determining the course of respiratory distress syndrome in premature infants and indicate that furosemide administration is beneficial when spontaneous diuresis does not occur. Furosemide may be particularly effective if combined with early closure of the ductus arteriosus.