To evaluate which of 4 established norms should be used for sonographic assessment of fetal femur length FL. Cross-sectional study using pooled data from 4 maternal-fetal medicine practices. Inclusion criteria were singleton fetus, gestational age 220/7 to 396/7 weeks, biometry measured, and fetal cardiac activity present. Four norms of FL were studied: Hadlock et al., the INTERGROWTH-21st Project (IG-21st), the World Health Organization Fetal Growth Curves (WHO), and the National Institutes of Child Health and Human Development Fetal Growth Studies unified standard (NICHD-U). The fit of our FL measurements to each norm was assessed by these criteria: mean z-score close to 0, standard deviation (SD) of z close to 1, low Kolmogorov-Smirnov D-statistic, high Youden J-statistic, close to 5% of exams <5th percentile, close to 5% of exams >95th percentile. In 26,177 ultrasound exams, our FL measurements had the best fit to the WHO standard (mean z-score 0.15, SD of z 1.02, D-statistic <0.01, J-statistic 0.95, 3.4% of exams <5th percentile, 7.0% of exams >95th percentile). The mean of the IG-21st standard was smaller than the other norms and smaller than our measurements, resulting in under-diagnosis of short FL. The mean of the Hadlock reference was larger than the other norms and larger than our measurements, resulting in over-diagnosis of short FL. The SD of the NICHD-U standard was larger than the other norms and larger than our observations, resulting in underdiagnosis of both short and long FL. Restricting the analysis to a subgroup of 7,144 low-risk patients without risk factors for large- or small-for-gestational age produced similar results. Of the norms studied, the WHO standard is likely best for diagnosis of abnormal FL.