Objective To evaluate salivary progesterone as a predictor of early preterm birth (PTB) and compare it with transvaginal sonographic (TVS) cervical length in asymptomatic high‐risk women. Design Prospective study. Setting Departments of Obstetrics and Gynaecology and Biochemistry at UCMS & GTBH, Delhi, India. Sample Ninety pregnant women. Methods The progesterone concentration in saliva of asymptomatic pregnant women at high risk for preterm delivery was estimated by immunoassay, and cervical length was measured by TVS, at the first antenatal visit at 24–28 weeks of gestation, and then repeated 3–4 weeks later. Main outcome measures Early PTB, mean and critical cut‐off values of salivary progesterone, and a diagnostic value comparison of salivary progesterone with TVS cervical length. Results The mean value of salivary progesterone was significantly lower in all women who delivered at <37 weeks of gestation (n = 38), compared with the term group (n = 52; P < 0.001). Salivary progesterone decreased significantly from the first to the second visit, with the maximum decrease observed in women who delivered at <34 weeks of gestation (29.6%, 95% CI 17.8–41.4%, P < 0.002). The single predictive critical cut‐off value for salivary progesterone was 2575 pg/ml, below which more than 80% of women delivered prematurely before 34 weeks of gestation, with sensitivity, specificity, and positive and negative predictive values of 83% (95% CI 58.6–96.4%), 86% (95% CI 75.9–93.1%), 60% (95% CI 38.6–78.8%) and 95% (95% CI 87.1–99.0%), respectively. The TVS cervical length decreased significantly (P < 0.001) in the women who delivered prematurely. Conclusions Low salivary progesterone concentration can be used for predicting early PTB in asymptomatic high‐risk women.