Preterm birth is the major clinical problem associated with perinatal mortality, serious neonatal morbidity and moderate to severe childhood disability in prosperous countries. Its prevalence is affected by the way in which gestational age is assessed, by national differences in the registration of births, associated practices, such as burial costs, or maternity benefits, which encourage or discourage registration, and by the perceived viability of extremely preterm infants. Despite these uncertainties, there is reliable evidence that preterm births are increasing, especially births before 28 weeks gestation. Contributing factors include births following assisted reproductive therapy and ovulation induction, especially multiple births, and the increasing proportion of births among women >34 years. On the other hand, improvements in neonatal care have substantially increased the survival of preterm infants during the last 15 years. There is wider acceptance of the importance of infection as a factor in preterm birth, and increasing recognition that processes leading to preterm birth may be initiated in very early pregnancy (the initiation of pre-eclampsia, major birth defects, premature placental separation), or even prior to pregnancy (prior pregnancy losses). It is unclear whether the familiar clinical presentations of preterm labour and birth reflect different pathophysiological processes. The pathways which link those processes to the consistent pattern of social differences in the probability of preterm birth have prompted new research approaches but in 2002 'the stubborn challenge of preterm birth' remains just that.