We conducted a multicenter randomized clinical trial to compare the efficacy and safety of high-frequency ventilation with that of conventional mechanical ventilation in the treatment of respiratory failure in preterm infants. Of 673 preterm infants weighing between 750 and 2000 g, 346 were assigned to receive conventional mechanical ventilation and 327 to receive high-frequency oscillatory ventilation. The incidence of bronchopulmonary dysplasia was similar in the two groups (high-frequency ventilation, 40 percent; conventional mechanical ventilation, 41 percent; P = 0.79). High-frequency ventilation did not reduce mortality (18 percent, vs. 17 percent with conventional ventilation; P = 0.73) or the level of ventilatory support during the first 28 days. The crossover rate from high-frequency ventilation to conventional mechanical ventilation was greater than the crossover rate from mechanical to high-frequency ventilation (26 vs. 17 percent; P = 0.01). High-frequency ventilation, as compared with conventional mechanical ventilation, was associated with an increased incidence of pneumoperitoneum of pulmonary origin (3 vs. 1 percent; P = 0.05), grades 3 and 4 intracranial hemorrhage (26 vs. 18 percent; P = 0.02), and periventricular leukomalacia (12 vs. 7 percent; P = 0.05). These results suggest that high-frequency oscillatory ventilation, as used in this trial, does not offer any advantage over conventional mechanical ventilation in the treatment of respiratory failure in preterm infants, and it may be associated with undesirable side effects.