Objective: This study’s objective was to determine whether the concentrations of β-human chorionicgonadotropin in the secretions of the cervix and vagina could be used to predict preterm delivery in a group of women at high risk for this complication. Study Design: Women attending a prematurity prevention clinic at an inner-city hospital July 1, 1996–October 1, 1997, were invited to participate. From those who consented, secretions from the cervix and posterior vaginal fornix were sampled every 2 weeks until delivery, beginning at 24 weeks’ gestation. Concentrations of β-human chorionic gonadotropin were measured with a commercially available enzyme-linked immunosorbent assay. Providers of obstetric care were blinded to the results. Levels of β-human chorionic gonadotropin in those who were delivered before 34 weeks’ gestation and those who were delivered at term were compared. A value >50 mIU/mL was considered elevated. This cutoff value was determined according to β-human chorionic gonadotropin values obtained during pregnancies that were delivered at term. Results: Of the 146 women asked to participate, 77 consented. There was no difference between participants and nonparticipants with respect to age, race, indication for enrollment in the clinic, gestational age at delivery, or parity. Of the 77 participants, 24 (31%) were delivered before 37 weeks’ gestation and 12 (16%) were delivered before 34 weeks’ gestation. A single β-human chorionic gonadotropin value >50 mIU/mL obtained between 24 and 28 weeks’ gestation was associated with a significant increase in the incidence of delivery before 34 weeks’ gestation (P = .03). This cutoff value had sensitivity, specificity, and positive and negative predictive values for predicting delivery before 34 weeks’ gestation of 50%, 87%, 33%, and 93%, respectively. Conclusion: These data suggest that the concentration of β-human chorionic gonadotropin in cervicovaginal secretions may be a useful predictor of preterm delivery.(Am J Obstet Gynecol 1998;179:870-3.)