A large cohort of consecutive live births with gestational ages assessed antenatally from 24 to 28 weeks from one tertiary center was studied to determine the association between mode of delivery and in-hospital mortality and morbidity and morbidity at two years of age. Between 1 January 1977 and 31 March 1982, 52.8% (172 of 326) of such infants survived their primary hospitalization. Obstetric factors independently associated with improved survival were increasing gestational age (P less than .0001), the absence of maternal hypertension (P = .007), singleton pregnancy (P = .007), and antenatal steroid therapy (P = .018). Although 62.7% (32 of 51) of infants delivered by cesarean section survived compared with 50.9% (140 of 275) of infants delivered vaginally, the increased survival was not statistically significant (X 2 = 1.97). Moreover, the trend favoring cesarean section disappeared after adjustment for confounding obstetric factors. In univariate analyses cesarean births more frequently required ventilatory support and there was a trend toward a lower incidence of cerebroventricular hemorrhage; again, however, when adjusted for extraneous factors these associations disappeared. Of the 172 in-hospital survivors, five died at home unexpectedly; 162 of the remaining 167 were traced; 18 (11.1%) had cerebral palsy and two (1.2%) were deaf. Of the 111 children who were fully assessed, 13.5% had major handicaps, 23.4% were suspect, and 63.1% were free of handicap at two years' corrected age. There was no association between mode of delivery and frequency of handicap. Little evidence was found from mortality or morbidity data to support routine delivery of infants of borderline viability by cesarean section.