This study investigates the accuracy of the postnatal gestational age assessment method of Ballard et al. to predict very preterm, preterm, term, postterm, and small-for-gestational-age infants, with gestational age by early ultrasonography used as the gold standard. Sensitivity, specificity, and predictive value were examined for each gestational age category, with 4193 single live births occurring during the period from 1982 through 1989 with an ultrasonographic examination done by 20 weeks' gestation and a postnatal assessment at 28 to 44 weeks. These data indicate that the postnatal assessment overestimated the gestational age of preterm births, resulting in an underestimation of very preterm and preterm percentages. Nearly three fourths of the postterm births by ultrasonography were misclassified by the method of Ballard et al. because of underestimation of postterm gestational ages. These data indicate that, for research purposes, appreciable misclassification of preterm, postterm, and small-for-gestational-age infants will occur if gestational age is determined by the technique of Ballard et al.