OBJECTIVES: Preterm premature rupture of fetal membranes (PPROM) exposes the fetus to preterm birth, and optimal timing for delivery is controversial. The aim of this study was to compare intentional early delivery ("active management") with expectant management in very preterm birth (28-32 weeks). STUDY DESIGN: We conducted a prospective randomized controlled trial with intent-to-treat analysis, at 19 tertiary-care hospitals in France and 1 in Geneva, Switzerland. Inclusion criteria were women age \textgreater/=18 years, PPROM at 28(0/7) to 31(6/7) weeks' gestation, singleton pregnancy. Exclusion criteria were maternal/fetal indications for immediate delivery. All participants received prophylactic antibiotics (amoxicillin + gentamicin) and two doses of corticosteroids. Women in expectant management delivered at 34 weeks, sooner if medically indicated. Women in active management delivered 24 h after the second steroid dose. The primary outcome measure was a composite of neonatal death/severe adverse events: periventricular leukomalacia, intraventricular hemorrhage, sepsis, oxygen requirement at 36 weeks, and necrotizing enterocolitis. The secondary outcome was clinical chorioamnionitis. RESULTS: The trial was stopped prematurely, due to recruitment difficulties. Of 360 women assessed, 139 (40% of calculated sample size) were randomized: 70 to expectant management, 69 to active management. Mean gestational age at PPROM was similar in both groups (30 +/- 1.3 vs. 30.2 +/- 1.2 weeks, respectively). There were 35 cases of medical/suspected complications requiring delivery in expectant management vs. 4 in active management. Mean latency between PPROM and delivery was 11.7 +/- 9.8 vs. 2.8 +/- 0.6 days, respectively; P \textless 0.0001 (median 8.4 (1.8-44.2) vs. 2.7 (1.9-4.3)). There were more caesarean deliveries in active than expectant management (80% vs. 60%, respectively; P \textless 0.01). There were 2 chorioamnionitis cases, both in expectant management. One baby died in expectant management; 2 in active management (one with heart defect). There was no significant difference in sepsis rates. The combined neonatal death/severe adverse events measure was 12.9% for expectant management and 13.0% for active management (OR 0.98; 95% CI: 0.33-2.93, P = 0.97). CONCLUSION: For PPROM at 28-32 weeks, and with antenatal antibiotic and steroid therapy, there were no observed differences in neonatal health when comparing expectant management to early delivery. As expected, expectant management resulted in higher gestational age and birth weight. However, our study was underpowered to draw firm and reliable conclusions.