To examine the relationship between the first-trimester serum biomarker levels (pregnancy-associated plasma protein A:PAPP-A; and free beta-human chorionic gonadotropin: b-hCG) and preterm birth; and to create the predictive models for preterm birth in case of strong correlation. Secondary analysis on a large prospective database of singleton pregnancies undergoing first-trimester serum screening with complete follow-up for pregnancy outcomes. The multiples of medians (MoM) of the biomarkers were compared between the group of term and preterm/early preterm birth. Predictive models were developed based on adjusted MoMs and logistic regression analysis, and then diagnostic performances in predicting preterm birth were assessed. Of 24,611 pregnancies eligible for analysis, 1908 (7.8%) and 500 (2.0%) had preterm and early preterm birth, respectively. Medians MoMs of both biomarkers were significantly lower in preterm and early preterm birth group. The predictive models were constructed. Performance in predicting preterm birth of these models yielded the area-under-ROC-curve of 0.560, 0.652, and 0.653 for b-hCG, PAPP-A, and combined biomarkers, respectively. In predicting early preterm birth, the areas-under-the-curve were found to be 0.551, 0.675, and 0.674 for b-hCG, PAPP-A, and combined biomarkers, respectively. The routine first-trimester serum screening of fetal Down syndrome could also be used as a tool of risk identification of preterm birth. We could take advantage of the screening by incorporating the predictive models into the Down syndrome screening software to report the preterm risk in the same test without extra effort and extra cost.