Objective: To evaluate the predictive value of thoracic fluid content (TFC) for weaning outcome in mechanically ventilated patients. Methods: In a prospective single-center observational study, 71 ventilated patients meeting weaning criteria were enrolled in the Emergency Intensive Care Unit (EICU) of Zigong Fourth People’s Hospital from April 2024 to May 2025. Prior to extubation, TFC, stroke volume, cardiac output, systemic vascular resistance and myocardial contractility were obtained non-invasively; arterial blood gas, B-type natriuretic peptide (BNP) and echocardiographic left-ventricular ejection fraction were also collected to calculate the oxygenation index. Patients were classified as weaning success (n = 48) or failure (n = 23) according to the need for re-ventilation within 48 h after extubation. ROC curves were constructed with R software to compare the predictive performance of TFC, BNP and oxygenation index for weaning failure. Results: Preliminary validation through analysis of clinical data revealed that the APACHE-II score was significantly higher in the failure group than in the success group (24.13 ± 7.84 vs. 18.17 ± 7.54, P = 0.004), and levels of BNP and TFC were also significantly elevated (P < 0.01). The AUC for predicting weaning failure using TFC was 0.927 (95% CI 0.861–0.993), superior to that of BNP (AUC 0.709) and the oxygenation index (AUC 0.519). When TFC was set at the optimal cutoff of 47 kΩ?1, the sensitivity was 85.2% and the specificity was 87.0%. Conclusion: TFC is an effective, non-invasive and bedside-reproducible