BACKGROUND: The INTERGROWTH-21st estimated fetal weight percentiles are recommended for predicting extremes of birthweight percentile, although evidence for their superiority over the widely employed Hadlock method is lacking. OBJECTIVE: This study aimed to compare the ability of estimated fetal weight percentiles calculated using the Hadlock method and the INTERGROWTH-21st method to predict extremes of birthweight percentile. STUDY DESIGN: A prospective cohort study of blinded serial ultrasonography in nulliparous women with a singleton pregnancy, The Pregnancy Outcome Prediction study was conducted in Cambridge, United Kingdom. The study participants who had a research estimated fetal weight performed at 36 weeks' gestation were eligible for the analysis. Estimated fetal weight percentiles for gestational age calculated using (1) the Hadlock method or (2) the INTERGROWTH-21st method were used as exposures. Birthweight percentiles of <10th (small for gestational age) and >90th (large for gestational age) for gestational age and fetal sex using the United Kingdom 1990 reference or the INTERGROWTH-21st birthweight reference were analyzed as outcomes using receiver operating characteristic curve analysis. Screening statistics from 2×2 tables were calculated for dichotomized exposures and each outcome. RESULTS: The Hadlock estimated fetal weight percentile performed better than the INTERGROWTH-21st estimated fetal weight percentile at discriminating both small for gestational age birthweight (areas under the receiver operating characteristic curves, 0.87 vs 0.85; 95% confidence intervals, 0.85-0.89 vs 0.83-0.87, respectively; P<.0001) and large for gestational age birthweight (areas under the receiver operating characteristic curves, 0.87 vs 0.86; 95% confidence intervals, 0.85-0.90 vs 0.83-0.89, respectively; P=.005). When the estimated fetal weight percentiles were dichotomized and screen positive was defined at 90% specificity, the sensitivity for the Hadlock vs the INTERGROWTH-21st method was 58.6% vs 52.3%, respectively, for small for gestational age, and 71.0% vs 60.9%, respectively, for birthweight at less than the third percentile (United Kingdom 1990 reference). The results were similar when the birthweight percentile was defined using the INTERGROWTH-21st birthweight reference, when the estimated fetal weight was calculated without the inclusion of head measurements, or when the women who had clinically indicated scans and women who had their research scan result were excluded. CONCLUSION: Replacing the Hadlock method by the INTERGROWTH-21st method may lead to less effective screening for extremes of birthweight percentile.