BACKGROUND The impact of prenatal depression on pregnancy outcomes is largely unknown. METHODS We conducted a population-based prospective cohort study among pregnant women of the Kaiser Permanente Medical Care Program to examine the impact of prenatal depression on the risk of preterm delivery. We interviewed pregnant women in their early pregnancy. Women's depressive symptoms were ascertained using the standard Center for Epidemiological Studies Depression Scale (CESD). The presence of significant prenatal depressive symptoms and severe depressive symptoms was determined by CESD scores >/=16 and >/=22, respectively. RESULTS Among the 791 participants who answered CESD questions and delivered a live birth, after controlling for potential confounders using the Cox proportional hazard regression, women with CESD scores >/=16 had almost twice the risk of preterm delivery compared with women without depressive symptoms: adjusted hazard ratio (aHR) = 1.9, 95% confidence interval (CI) 1.0-3.7. The risk of preterm delivery increased with increasing severity of depression: aHR = 1.6 (CI 0.7-3.6) for CESD 16-21 and aHR = 2.2 (CI 1.1-4.7) for CESD >/=22. The risk of preterm delivery associated with prenatal depression appears to be exacerbated by low educational level, a history of fertility problems and the presence of obesity and stressful events. The observed associations were not confounded by the use of antidepressants, although some of the associations did not reach statistical significance. CONCLUSIONS Our findings show that pregnant women with depressive symptoms are at increased risk of preterm delivery and, in addition, provide preliminary evidence that social and reproductive risk factors as well as obesity and stressful events may exacerbate the effect.