OBJECTIVE We prospectively evaluated the accuracy of a gestational age-independent method of detecting abnormal growth, the transverse cerebellar diameter/abdominal circumference ratio, and compared this with standard ultrasonographic methods of growth assessment. METHODS We prospectively studied 825 low-risk obstetric patients and 250 patients having risk factors for fetal macrosomia (n = 92) or growth retardation (n = 158). Measured fetal parameters included the biparietal diameter, head circumference, transverse cerebellar diameter, abdominal circumference, and femur length. The estimated fetal weight, head circumference/abdominal circumference, cerebellar diameter/abdominal circumference, and femur length/abdominal circumference ratios were calculated. Reference curves for these parameters were created from a cross-sectional analysis of the low-risk group. Univariate analysis was used to determine the sensitivity, specificity, predictive values, and odds ratios of each individual parameter in identifying a small- or large-for-gestational-age infant. A multivariate logistic regression model with a variable selection procedure was then used to determine whether significance remained when we controlled for other parameters. RESULTS Within the low-risk group, the transverse cerebellar/abdominal circumference ratio was gestational age independent between 14 and 42 weeks with a mean of 13.68 +/- 0.96. A value exceeding 2 SD of the mean was significantly associated with birth or a small-for-gestational-age infant, being abnormal in 98% and 71% of asymmetrically and symmetrically growth-retarded infants, respectively. Significance was maintained in the multivariate regression model. The ratio was not helpful in detecting the large-for-gestational-age infant. CONCLUSIONS The fetal transverse cerebellar diameter/abdominal circumference ratio is an accurate, gestational age-independent method of identifying the small-for-gestational-age but not the large-for-gestational-age infant.