Most industrialized countries have reported recent decreases in the size of infants born at term but no reduction, or even a rise in the incidence of preterm birth. This paper reviews recent secular trends in preterm birth and the evidence about possible reasons for those trends. The hypothesized reasons include ignorance about the causal determinants of preterm birth, failure to reduce exposure to recognized determinants, increases in multiple births, the use of early ultrasound (rather than menstrual dates) to estimate gestational age, early induction or cesarean section for pregnancy complications, and registration of extremely immature births of borderline viability.