<p>A comprehensive comparative analysis of machine learning algorithms, for predicting the Pavement Condition Index (PCI) of rigid pavement is presented in this study. A three-year longitudinal dataset (2021–2023), which of comprises 42 distress indicators derived from 14 distress types evaluated at three severity levels was used as the feature space for model development. Five Machine learning (ML) PCI prediction models, Linear Regression (LR), Artificial Neural Networks (ANN), K-Nearest Neighbors (KNN), Support Vector Regression (SVR), and Random Forest (RF) were trained and evaluated. SVR performed best in terms of highest accuracy (R² = 0.98) and ability to model nonlinear patterns of deterioration and this model was also the best of all models at capturing the changes in deterioration pattern. When using the feature importance analysis, the feature with the highest influence was Multiple Cracks Intersecting for the High-severity distresses. Temporal validation confirmed predictive robustness using 2021 data to predict 2022 conditions (R² = 0.89) and combined 2021–2022 data for the 2023 predictions (R² = 0.94) which demonstrates year-ahead forecasting capability. This study provides valuable data for pavement management systems and infrastructure maintenance planning for more accurate pavement condition forecasting and better resource allocation.</p>