OBJECTIVE: Our purpose was to assess the accuracy of third-trimester ultrasonographic biometry in the diagnosis of small-for-gestational age fetuses in a low-risk obstetric population. STUDY DESIGN: A total of 1000 low-risk pregnancies were prospectively examined at 31 weeks' gestation. The diagnostic accuracy of the fetal abdominal circumference and estimated fetal weight according to 24 formulas in the literature were evaluated by the use of receiver-operator characteristic curves. RESULTS: The incidence of a birth weight <10th percentile was 8.2%. The diagnostic accuracies of abdominal circumference and estimated fetal weight were remarkably similar. None of the 24 formulas performed significantly better than the measurement of the abdominal circumference. At a specificity of 90%, 46% of infants with a birth weight <10th percentile and five of six cases with adverse perinatal outcomes were predicted. CONCLUSION: In a low-risk obstetric population third-trimester ultrasonographic biometry has limited value in predicting small-for-gestational-age fetuses, and estimation of fetal weight does not carry an advantage over measurement of the abdominal circumference. The optimal cutoff value remains uncertain. However, by accepting a 10% false-positive rate, half of small-for-gestational-age fetuses and most perinatal complications could be recognized. (AM J OBSTET GYNECOL 1996;174:1037-42.)