Preeclampsia is characterized by diffuse vascular endothelial dysfunction. Tumor necrosis factor-α (TNF-α), which plays a key role in the cytokine network responsible for immunoregulation, is also known to contribute to endothelial dysfunction and other metabolic disturbances noted in preeclampsia. Results from cross-sectional studies and one longitudinal study indicate that TNF-α (or its soluble receptor, sTNFp55) is increased in the peripheral circulation and amniotic fluid of women with preeclampsia as compared with normotensive women. Between December 1993 and August 1994, prediagnostic sTNFp55 concentrations (a marker of excessive TNF-α release) were measured in 35 women with preeclampsia and 222 normotensive women to determine whether elevations precede the clinical manifestation of the disorder. Logistic regression procedures were used to calculate maximum likelihood estimates of odds ratios and 95% confidence intervals. Mean second trimester (15–22 weeks' gestation) serum sTNFp55 concentrations, measured by enzyme-linked immunosorbent assay, were 14.4% higher in preeclamptic women than in normotensive controls (716.6 pg/ml (standard deviation 193.6) vs. 626.4 pg/ml (standard deviation 158.0); p = 0.003). The relative risk of preeclampsia increased across successively higher quintiles of sTNFp55 (odds ratios were 1.0, 1.3, 2.1, and 3.7, with the lowest quintile used as the referent; p for trend = 0.007). After adjustment for maternal age, adiposity, and parity, the relative risk between extreme quintiles was 3.3 (95% confidence interval 0.8–13.4). These findings indicate that the level of TNF-α in maternal circulation is increased prior to the clinical manifestation of the disorder, and they are consistent with the hypothesized role of cytokines in mediating endothelial dysfunction and the pathogenesis of preeclampsia. Further work is needed to identify modifiable risk factors for the excessive synthesis and release of TNF-α in pregnancy, and to assess whether lowering of TNF-α concentrations in pregnancy alters the incidence and severity of preeclampsia. Am J Epidemiol 1999; 149: 323–9.