We model unique state interventions to rescue commercial banks during the 2008-09 global financial crisis with the complementary binary logistic models that accommodate their skewed distribution. Our findings show that large and illiquid banks, and banks from countries with weak regulations, and weak shareholder and creditor rights are more likely to receive state intervention. These findings remain robust to a restricted definition of state intervention, alternative measures of bank fundamentals, placebo estimations, counterfactual sampling with propensity scores, and bank and country sample splits. These bank and incremental country level predictors can help regulators and supervisors limit future state interventions.