Analysis of singleton preterm birth from vital statistics data is hamp
ered by inaccurate gestational age dating. This is most notably eviden
ced by the large fraction of implausibly high birthweights for gestati
onal age given in birth records. Using natality statistics for New Eng
land, 1977-88, birthweight distributions were plotted for representati
ve preterm gestations. Confounding by improbable birthweights was obse
rved to be considerably less among preterm twin births than among pret
erm singleton births.