Résumé : Purpose: To prospectively define fetal density in the second half of pregnancy by using magnetic resonance (MR) imaging and to compare estimates of fetal weight based on ultrasonography (US) and MR imaging with actual birth weight. Materials and Methods: Written informed consent was obtained for this ethics committee - approved study. In this cross-sectional study between March 2011 and May 2012, fetal density was calculated as actual birth weight at delivery divided by fetal body volume at MR imaging in 188 fetuses between 20 weeks and 2 days and 42 weeks and 1 day of gestational age. Regression analysis was used to investigate the effect of variables, including sex, on fetal density. The US estimate of fetal weight was performed according to Hadlock et al, and the MR estimate of fetal weight was calculated based on the equation developed by Baker et al. US and MR estimates of fetal weight were compared with actual birth weights by using regression analysis. Results: Median fetal density was equal to 1.04 (range, 0.95-1.18). Fetal density was significantly associated with gestational age at delivery but not with fetal sex. In 26.6% of fetuses, the US estimate of fetal weight had a relative error of more than 10%, while a similar relative error for the MR estimate of fetal weight occurred in only 1.1% of fetuses. The limits of agreement were narrower with the MR estimate of fetal weight compared with the US estimate of fetal weight. Conclusion: In the second half of pregnancy, fetal density varies with gestational age. Fetal weight estimates by using fetal MR imaging are better than those by using prenatal US. © RSNA, 2013.