Despite the popular current distinction between "proportional" and "disproportional" intrauterine growth retardation, it has never been shown that variation in body proportions is greater among growth-retarded than nongrowth-retarded infants of the same birth weight, nor that proportionality is distributed bimodally among growth-retarded infants. Based on a cohort of 8719 neonates born between 1980 and 1986 of mothers with concordant (+/- 7 days) menstrual dating and early ultrasound estimates of gestational age, we used a continuous measure of birth weight for gestational age to define four study groups: nongrowth retarded (n = 5163) and mild (n = 411), moderate (n = 226), and severe (n = 147) intrauterine growth retardation. Compared with non-growth-retarded infants of the same gestational age, growth-retarded infants had substantially lower lengths, head circumferences, and proportionality ratios, and the magnitude of the deficits increased significantly with increasing degrees of growth retardation. When the comparison was based on birth weight rather than gestational age, however, growth-retarded infants had slightly but significantly greater lengths and head circumferences, with increased variability in body proportions, but no evidence of the bimodality that would characterize two distinct subtypes. The analysis suggests that proportionality among intrauterine growth-retarded infants represents a continuum, with progressive disproportionality as severity of growth retardation increases. Moreover, despite evidence of some "sparing," the absolute magnitudes of the deficits in length and head growth remain substantial.