In Reply.— We appreciate the opportunity to respond to the thoughtful criticisms by Lieberman et al. These criticisms involve our interpretation of the literature, the age of the Collaborative Perinatal Project, potential nonindependence of the observations, and the possibility that women who have repeated hematocrits are different from the general population. We address each of these concerns. Our purpose in citing Nilsen et al 1 was to provide supporting evidence for our contention that hematocrit rises during the third trimester of a normal pregnancy, not to address the relationship between anemia and preterm birth. These investigators demonstrated that in normal pregnancies to nonsmoking, nonanemic women, the hemoglobin level rose approximately 5 g/L between 25 to 30 weeks and term. Taylor and Lind, 2 who studied unselected British women, noted that among those who received iron, the mean hematocrit increased from 33 percent at 26 to 30 weeks to 36 percent at