OBJECTIVE: Identifying patients at risk for preterm delivery continues to be difficult. We analyzed fetal fibronectin immunoassays in 763 subjects with threatened preterm delivery from 10 clinical sites. STUDY DESIGN: Symptomatic women between 24 weeks 0 days and 34 weeks 6 days with intact amniotic membranes and cervical dilatation <3 cm were included. Specimens were obtained from the posterior vaginal fornix and concentrations of fetal fibronectin were measured by enzyme-linked immunosorbent assay. A total of 150 (20%) patients had positive results for the fetal fibronectin immunoassay. The simultaneous effects of multiple variables on predicting positive fetal fibronectin immunoassay results were analyzed with use of logistic regression models. Variables included were patient age (mean 24.2 years), race (40% white), gravidity (29% primigravid), history of previous premature infants (15%), sexual activity within 24 hours of sample collection (66/763 or 9%), cervical examination within 24 hours of sample collection (107/763 or 14%), vaginal bleeding (118/759 or 16%), uterine contractions (192/750 or 26% with three or more in 1 hour), cervical dilatation (94/763 or 12% with dilatation between 1 and 3 cm), estimated gestational age at sampling (mean 30 weeks 2 days). RESULTS: Regardless of the variable selection process (e.g., forward, backward, or stepwise), the reduced model showed five significant variables in predicting positive fetal fibronectin: cervical dilatation (χ 2 33.5, p = 0.0006), sexual activity within 24 hours of sample collection (χ 2 22.9, p < 0.0001), presence of vaginal bleeding (χ 2 17.0, p < 0.0001), cervical examination within 24 hours of sample collection (χ 2 11.6, p < 0.0006), and uterine contractions (χ 2 6.8, p = 0.01). Deletion of the true positive subset did not change these variables or the magnitude of risk. CONCLUSION: Cervical dilation or manipulation predicts positive fetal fibronectin assays and may serve to explain some of the false-positive results generated by the test. (Am J Obstet Gynecol 1997;176:639-41.)