This study investigates the performance of the Random Forest algorithm in measuring the quality of Higher Education Institutions (HEIs) in Indonesia. The current reliance on administrative evaluations and conventional accreditation processes often fails to capture the institutions’ actual performance comprehensively, indicating the need for a data-driven alternative. This research proposes the use of a Random Forest–based classification model to assess institutional quality based on relevant accreditation indicators. The RF-D model demonstrates optimal classification performance across three quality categories—Good, Very Good, and Excellent—with high precision, recall, and F1-scores for all classes. The Very Good category achieves a precision of 89% and a recall of 80%, while the Excellent category records the highest recall at 86%. Furthermore, the Area Under Curve (AUC) scores, which approach 1.0 for all categories, confirm the strong discriminative capability of the model. This study also highlights the influence of train–test data ratios on model stability. Extreme imbalances in data splitting can lead to overfitting or underfitting, emphasizing the importance of selecting an appropriate configuration during model development. Overall, the findings indicate that Random Forest, when optimized with suitable parameters, provides a more accurate, objective, and replicable approach for evaluating HEI quality in Indonesia. These results are expected to contribute to the development of a more transparent higher education assessment system and support data-driven decision-making among policymakers.