OBJECTIVE Our purpose was to assess the utility of cervicovaginal expression of fetal fibronectin in the diagnosis of preterm labor. METHODS Women seen between 24 and 34 weeks' gestation with symptoms of preterm labor, intact membranes, and cervical dilatation < 3 cm were enrolled at five university medical centers. Cervicovaginal swabs were obtained and assayed for the presence of fetal fibronectin by means of a monoclonal antibody assay. Results were compared with cervical dilatation and uterine contraction frequency as indicators of interval to delivery and delivery before 37 weeks. RESULTS A total of 192 eligible women at a mean gestational age of 30.8 +/- 2.9 weeks were enrolled from a population of 418 subjects screened. The rate of preterm birth was 32.3% (62/192). The mean interval from presentation to delivery was 25.3 +/- 24.1 days in the 45 subjects with a positive fibronectin assay and 52.4 +/- 24.8 days in the 147 subjects with a negative assay (p = 0.0001). The sensitivity, specificity, and positive and negative predictive values of fetal fibronectin expression for delivery < 37 weeks were 44% (27/62), 86% (112/130), 60% (27/45), and 76% (112/147). The fetal fibronectin assay was especially useful in predicting risk of delivery within 7 days (sensitivity 93% [13/14], specificity 82% [146/178], positive predictive value 29% [13/45], and negative predictive value 99% [146/147]) and was notably superior to both cervical dilatation > 1 cm and contraction frequency greater than or equal to eight per hour (sensitivities 29% and 42%, specificities 82% and 67%, positive predictive values 11% and 9%, and negative predictive values 94% and 94%, respectively. CONCLUSIONS Cervicovaginal fetal fibronectin predicts delivery within 7 days more accurately than do cervical dilatation and contraction frequency in a population of women evaluated for early preterm labor.