Head and abdominal circumferences (HC, AC) can be measured directly or estimated from diameters. Accurate prediction of circumference growth after 26 weeks depends on growth model specification and the quality of the circumference measurement method used. To evaluate the effect of the method of measurement, direct and estimated HC and AC values from 20 normal fetuses, obtained between 15 and 38 weeks, were used to determine two pairs of functions for specifying HC and AC growth models. The two HC and AC data sets and pairs of model specification functions were used in various combinations to determine HC and AC growth models, to predict growth after 26 weeks, and to compare predicted values to actual measurements. Systematic and random prediction errors were determined for each combination. For HC no differences in either type of error were found. Similar results were obtained for AC with one exception, a statistically significant difference in systematic error between the directly measured circumferences - model specification based on estimated values combination and the estimated circumferences - model specification based on directly measured values combination. These results indicate that the prediction of future HC and AC growth is not significantly affected by the type of circumference measurement procedure used except in extreme cases where circumference measurements are made on parts of the body that can be easily distorted.