OBJECTIVES: End-of-life decisions in intensive care units (ICUs) are complex, influenced by patient severity, treatment efficacy and resource constraints. This study assessed the predictive value of delta-Sequential Organ Failure Assessment (SOFA) scores on days 2, 3 and 5, both independently and combined with Simplified Acute Physiology Score (SAPS) 3, for in-hospital mortality. METHODS: A retrospective cohort study analysed ICU patients with stays of≥5 days from 2018 to 2020. Clinical data included SAPS 3 and SOFA scores at admission, 48 hours and on days 3 and 5, alongside mortality outcomes. Logistic regression and discriminant analysis identified mortality predictors and evaluated model performance. RESULTS: Among 200 patients, 26% died in hospital. Non-survivors had significantly higher SAPS 3 scores (mean 51.9±11.9 vs 45.6±11.9 in survivors, p=0.001) and worsening SOFA scores, particularly on days 3 and 5. Combining SAPS 3 with delta-SOFA on day 5 produced an area under the receiver operating characteristic (AUROC) of 0.71 (95% CI: 0.63 to 0.79), indicating moderate predictive ability. CONCLUSIONS: Integrating the delta-SOFA score on day 5 with SAPS 3 improves in-hospital mortality predictions during prolonged ICU stays. Findings suggest early treatment limitations within 48 hours may be premature, as patient responses evolve and shared decision-making becomes more feasible over time.