Predicting financial distress is crucial for firms, investors, and policymakers to manage risks and maintain financial stability. This study applies machine learning to predict financial distress among Malaysian non-financial firms using data from 2015 to 2023. Among the models tested, Random Forest proves the most reliable, consistently outperforming other classifiers across training, validation, and testing. With optimised hyperparameters, the model ensures high accuracy and generalisability, detecting distress patterns with minimal errors. This reinforces the significance of machine learning in early detection and risk assessment while validating Random Forest’s reliability in predicting financial distress.