Data on structural chromosome abnormalities identified during prenatal diagnosis were used to estimate the number of such abnormalities that would be detectable in an unselected series of newborns using moderate levels of banding (400 to 500 bands). These estimates were compared with the rates detected in nonbanded surveys of newborns. Between 1976 and 1990 prenatal diagnosis using banding techniques was carried out in our laboratory on 14,677 women aged 35 and over. Among these, we detected 112 structural rearrangements, 32 unbalanced and 80 balanced. These figures were adjusted by two methods to give an estimate of the frequency of structural abnormalities in the newborn. Our data suggest that the use of moderate levels of banding increases the frequency of unbalanced structural abnormalities from 0.052 to 0.061% and of balanced structural abnormalities from 0.212 to 0.522%. Thus, the total number of chromosome abnormalities detectable in the newborn is increased from 0.60% in unbanded preparations to 0.92% in banded preparations.