One hundred and eight-six pregnant women were studied with ultrasound for clinical suspicion of poor fetal growth. Fetal weight was estimated using biparietal diameter and mean abdominal diameter with a special nomogram. The fetal femur length (FL) to abdominal circumference (AC) ratio was also calculated. The mean interval between the last ultrasound examination and delivery was 9 days (range, 0 to 14 days). Eighty-three women had their last ultrasonic examination 0 to 4 days (mean, 2 days) before delivery. In this group the arithmetic mean of weight estimation errors was -1 gm (SD, 159 gm). The signed mean percent error was +0.6% (SD, 6.6%). Neither systematic nor random errors were found between different growth percentile groups. When small for gestational age (SGA) was defined as birthweight below 2.5 percentile, the sensitivity, specificity, and positive predictive value of the weight estimations were 82%, 92%, and 84%, respectively, in 186 cases. Significant differences were found in FL to AC ratios between growth pattern groups but fetal weight estimation was found to be superior in the detection of SGA fetuses.