Postnatal assessment of gestational age in preterm neonates traditionally has been performed using the methods of Dubowitz and Ballard. This study was designed to determine the accuracy of these methods in a sample of very low birth weight preterm neonates. Dubowitz and Ballard examinations were done on 110 preterm neonates within the first 72 hours of life by a neonatologist masked to the gestational age assessed antenatally. Mean birth weight was 1066 +/- 256 g (SD). These data were compared with gestational age assessments using last menstrual period and best obstetric estimate calculated by an obstetrician unaware of the neonatal examination. Mean gestational age using last menstrual period was 28.3 +/- 2.9 weeks. Mean differences between last menstrual period and Dubowitz/Ballard were -2.8 +/- 2.1 weeks and -2.6 +/- 2.2 weeks, respectively. Results using best obstetric estimate were similar. An ophthalmologist examined lens vessels of 89 neonates. A similar pattern toward overestimation of gestational age interval by Dubowitz/Ballard was seen at each lens vessel grade. The Dubowitz and Ballard examinations are inaccurate methods of assessing gestational age in preterm neonates with birth weights less than 1500 g.