This study aims to predict the financial failure of manufacturing sector firms traded on the Borsa İstanbul using tree-based machine learning methods and to identify the most successful method. Random Forest (RF) and Extreme Gradient Boosting (XGB) models were used in the prediction process, and financial failure was predicted for 204 firms using data from the 2023-2024 period. The prediction results showed that the RF model performed better, and the variable that contributed most to the prediction of the models was found to be the Operating Profit / Short-Term Liabilities ratio. The results demonstrate the performance of two different tree-based machine learning methods in predicting financial failure using data from the manufacturing sector and emphasize the importance of operating profitability as the capacity to pay short-term debts in predicting financial failure.