Physicians who deliver babies in Alabama were surveyed to determine their level of knowledge about the survival of low--birth weight/early gestational age infants. These estimates of neonatal survival were compared to the actual neonatal survival rates at local hospitals and at the regional perinatal centers in Alabama. In addition, the physicians' knowledge of survival rates was correlated with their management decisions in hypothetical cases of premature labor. Our findings indicate that physicians who perform deliveries tended to underestimate the potential for neonatal survival in premature infants. Equally as important, the range of responses varied markedly. In the hypothetical cases, management decisions often appeared to be based on incorrect information about neonatal survival. These decisions, including not electronically monitoring fetuses, not performing a cesarean section for fetal distress, and not transferring women in premature labor to a perinatal center, if made in actual cases, would result in potentially viable fetuses receiving less than optimal management.