Risk assessment in AML is mainly based on cytogenetic and molecular classification. Even with these systems, induction failure and early death remain common, especially in older patients. In clinical practice, patients often do poorly when aggressive leukemia biology and patient frailty occur together. We aimed to better define these high-risk patterns by identifying specific combinations of clinical and genomic factors linked to early treatment failure. We analyzed 942 specimens out of 805 patients from the BeatAML2 cohort and applied fault tree analysis (FTA), a structured method that identifies small groups of factors that, when present together, are associated with clearly higher event rates. Separate models were built for induction failure and 60-day mortality using gene mutations and routine clinical variables available at diagnosis. RUNX1 and TP53 mutations were strongly associated with failure to achieve remission, particularly when combined with low platelet counts. Early mortality was more closely linked to patient-related factors, especially advanced age together with low albumin or impaired kidney function. We defined an extreme-risk constellation (XRC_ANY) as the presence of at least one high-risk combination. This was observed in 32.6% of patients and was associated with significantly shorter overall survival (median 297 vs 861 days, p<0.0001). In multivariable Cox analysis, XRC_ANY remained independently associated with survival after adjustment for cytogenetic-molecular risk classification (hazard ratio 2.03, p<0.0001). Overall, these findings suggest that poor outcomes in AML often occur when aggressive leukemia and patient frailty are present at the same time. Fault tree analysis may help refine risk assessment within existing classification systems.