The aim of the present study was to determine whether mothers reporting more life events experience more preterm births following both complicated and uncomplicated pregnancies. A Life Events Inventory was administered prospectively to women at high risk for poor obstetric and neonatal outcomes who took part in the Pregnancy Home Visiting Program (PHVP), a randomized controlled trial of the effect of a programme of antenatal home visits by midwives on the incidence of preterm birth. This study took place in Western Australia in the years 1984-1987. All women in the study had had a previous poor pregnancy outcome. The women were classified into two groups--those with complicated and those with uncomplicated pregnancies. Pregnancies classified as 'complicated' were defined as a pregnancy in which there was antenatal hospital admission(s) for hypertension, antepartum haemorrhage or other medical reason except for preterm birth. Pregnancies classified as 'uncomplicated' refer to all pregnancies without these complications. No significant association was found between life events and preterm birth although the total stress score for women with uncomplicated pregnancies almost reached significance, as did the number of life events for both women with complicated and women with uncomplicated pregnancies considered together. Life events were not shown to have a predictive relationship to preterm birth even when stratified by etiologically different groups. However, although stress was not an important predictor of preterm birth in this group of women at biological risk it may yet be so in a group at social risk.