Purpose: In our study we aimed to predict preterm delivery using the salivary progesterone levels. Material and Method: In this follow-up study, we collected salivary samples from 100 pregnant women at 24, 27 and 30 weeks' gestation. Pregnant women with preeclampsia, diabetes, twin pregnancy and intrauterine growth restriction were not included in the study. The salivary progesterone levels were determined by enzyme immunoassay. The patients were followed until delivery. Findings: Preterm birth occurred in 11 (9.09%) of the 100 patients. The mean progesterone levels of patients that delivered preterm were lower than the levels of the term deliveries. Statistically, the progesterone levels were significant at week 24 and 27 (p=0.031, p=0.018). Although the progesterone levels of the preterm births at week thirty were low, this was not found to be statistically significant (p=0.061). The preterm birth rate in patients with no history of preterm birth was 2.2%, and in patients with a positive history of preterm birth the rate was 27.3% (p=0.004). In the ROC curve, with respect to predicting preterm birth the progesterone levels at week 27 were more significant than the levels at weeks 24 and 30. Result: Assessing the salivary progesterone level is a non-invasive method with high sensitivity, its assessment between 20-30 weeks of pregnancy is useful for predicting preterm birth. However, it must not be forgotten that preterm birth is a multifactorial process. Considering this, it is important to evaluate the markers and the clinical features togetherto predict and prevent preterm birth instead of using a single marker alone. By doing so, high-risk patients can be identified and the appropriate treatment can be delivered.