The goal of this study was to evaluate the influence of preeclampsia on preterm delivery, examining whether the association varied among preterm birth subtypes defined by gestational age and precipitating events. A population-based, longitudinal study of the association between mild and severe preeclampsia and preterm birth subtypes was conducted among 59,851 women (resulting in a total of 78,086 pregnancies) delivering singleton live births in the province of Nova Scotia, Canada between 1986 and 1992, utilizing the Nova Scotia Atlee perinatal database. Very preterm (< 33 weeks' gestation) and moderately preterm (33-36 weeks' gestation) births were further classified as occurring due to (1) membrane rupture, (2) medical intervention, and (3) spontaneous onset of labor (before membrane rupture). Mild and severe preeclampsia occurred in 8.7 and 1.7% of pregnancies, respectively, after exclusions of multiple births. After adjustment for confounders by multivariable logistic regression based on the generalized estimating equations, severe preeclampsia was strongly associated with the risk of very preterm birth (RR = 80.8, 95% CI: 54.2-120.6), and moderately preterm birth (RR = 41.8, 95% CI: 34.0-51.4) due to medical intervention. A less dramatically elevated risk of very preterm (RR = 2.1, 95% CI: 1.1-4.0) and moderately preterm (RR = 2.2, 95% CI: 1.7-2.9) birth due to medical intervention was apparent among pregnancies complicated by mild preeclampsia. Very preterm births due to membrane rupture were too rare to examine, but moderately preterm births due to membrane rupture were not associated with preeclampsia. Preeclampsia was associated with an increase in the risk of moderately preterm births due to spontaneous labor (RR = 1.9, 95% CI: 1.3-2.8), but not very preterm births (RR = 1.0, 95% CI: 0.7-1.2). Substantial variability was observed in the association between preeclampsia and preterm birth in relation to the subtypes defined by gestational age and pathway, with strong associations between hypertension and medically induced preterm births. The results indicate a need to separate preterm births into subcategories to properly evaluate the association between preeclampsia and preterm births and interventions to reduce the adverse effects of preeclampsia.