The accuracy of cervicovaginal fetal fibronectin as a predictor of preterm birth was studied in patients with increased risk for preterm delivery (according to the Creasy-score). In a prospective blind observational study the smear from the posterior fornix vaginae of 56 pregnant patients without PROM was examined using a quantitative immunoassay for the detection of fetal fibronectin. The patients who tested positively for fetal fibronectin had significantly more preterm deliveries than those with a negative result (CHI square-test, p < 0.01, RR 5.1). Overall, sensitivity, specificity, positive and negative predictive values were 56%, 87%, 45% and 91%, respectively. In patients with preterm labor these values were 75%, 87%, 60%, and 93%, respectively. No patient with a negative result delivered preterm during the following two weeks. It is concluded that performing the fetal fibronectin test in patients with preterm labor is useful for the prediction of preterm birth. Routine testing in patients at increased risk (asymptomatic patients) is not recommended for lack of effectiveness.