OBJECTIVE: Our purpose was to compare the predictive values for preterm delivery of fetal fibronectin and cervical length measured by transvaginal ultrasonography and to determine whether performing both tests improves their separate predictive values. STUDY DESIGN: This prospective blinded study performed both tests on 76 patients hospitalized with signs of premature labor between 24 and 34 weeks of gestation. The outcome measure was delivery before 37 weeks' gestation. RESULTS: The rate of preterm birth was 26.3% (20/76). The predictive values of fetal fibronectin and of a cervical length of ≤26 mm, considered separately, were approximately equal, and the negative predictive value of each was excellent (86.6% and 89.1%, respectively). This value improved slightly when positive fetal fibronectin, a cervical length ≤26 mm, or both defined abnormality (negative predictive value 94.4%). The positive predictive values, although less helpful, were still useful (45.2% and 50.0%, respectively). Combining both indicators did not noticeably improve the positive predictive value (52.4%). The risk of preterm delivery for a patient with a positive fetal fibronectin level and a short cervix was high (odds ratio 13.9, 95% confidence interval 3.7 to 52.2). CONCLUSION: Fetal fibronectin and cervical length are approximately equivalent in their ability to distinguish between patients at high and low risk for preterm delivery. For physicians equipped to perform transvaginal ultrasonography, however, the additional information about the fibronectin level provides only slight benefits. (Am J Obstet Gynecol 1997;176:196-9.)