The ability to estimate fetal weight accurately from symphysis-fundal height (SFH) and abdominal girth (AG) measurements was evaluated. Multivariate regression analysis was performed on measurements made on 208 primigravidas within 48 hours of delivery. A better correlation with birth-weight was obtained with the SFH than with the AG. The equation BW = -1.515 + 0.092 (SFH) + 0.016 (AG) yielded a coefficient of multiple correlation of 0.7259. Using this equation, the mean percentage prediction errors were 5.7% +/- 4.2(SD) in fetuses between 2,500 and 3,500 g (n = 156), 9.4% +/- 5.25(SD) in those larger than 3,500 g (n = 33), and 19.1% +/- 8.2(SD) in those smaller than 2,500 g (n = 19). All the generated equations similarly underestimated the fetal weight in the larger babies and overestimated in the smaller babies. Inclusion of quadratic and logarithmic functions as well as skin-fold thickness measurements into the equations did not decrease the error or alter the distribution of errors. It was concluded that although fetal weight estimation may be reasonably accurate between 2,500 g and 3,500 g, the error is too great for the method to be clinically useful in the smaller and larger babies. Moreover, a theoretical basis for a high sensitivity in the detection of small for dates fetuses from these parameters would be difficult to establish.