The used car sales business in Indonesia has been experiencing rapid growth, driven by increasing market demand. However, determining the price of used cars remains a challenge due to various influencing factors such as the year of production, mileage, and vehicle specifications. This research develops a web-based used car price prediction system using the XGBoost Regressor algorithm. The data used undergoes preprocessing and hyperparameter tuning to produce a high-performance model (R²: 97.79% on training data and 89.90% on testing data, MSE: 2.3129, RMSE: 1.5208). Additionally, the system provides a car recommendation feature using a Rule-Based Method, allowing users to filter vehicles based on specific criteria. The results demonstrate that this system effectively assists both buyers and sellers in making more informed, efficient, and transparent decisions in used car transactions.