In this study we used receiver-operating characteristic (ROC) analysis to comparatively evaluate maternal periodontal parameters to predict preterm (PB) delivery and low birth weight (LBW) delivery among Jordanian women. A total of 277 pregnant women (20 weeks of gestation or less) had periodontal examination at baseline and followed up until delivery. Gestational age and birth weight were retrieved from their medical records. ROC curve analyses were used to examine the overall discriminatory power of the studied periodontal parameters to predict PB, LBW, and PB or LBW. For the three outcome variables, the area under curve (AUC) ranged from 0.84 to 0.87 for average clinical attachment level (CAL), 0.78-0.86 for percent of sites with CAL ≥ 5 mm, 0.63-0.74 for percent of sites with CAL ≥ 6 mm, and 0.71-0.82 for number of missing teeth indicating that they had high discriminating power to predict adverse pregnancy outcomes. All other parameters had AUC less than 0.60 and thus had low discriminating power. Average CAL performed the best in predicting the studied adverse pregnancy outcomes because it has the highest AUC. The severity and extent of periodontal disease as measured by CAL can be used to predict the occurrence of adverse pregnancy outcomes.