Glycoprotein (GP) IIb/IIIa inhibitors are often used as a rescue or bailout therapy to manage complications arising during percutaneous coronary intervention, rather than as prophylactic treatment. We sought to identify the characteristics and outcomes of patients requiring bailout treatment. The ESPRIT trial randomized 2,064 patients to receive eptifibatide or placebo starting immediately before percutaneous coronary intervention (PCI). Bailout therapy was used in 77 patients: 43 (4.2%) randomized to placebo and 34 (3.3%) to eptifibatide (p = 0.3). Bailout therapy for thrombosis was used more often in the placebo group (2.1% versus 1.0%; p = 0.03). Multivariable predictors of bailout included a greater than or equal to 90% stenosis, or visible thrombus on the baseline angiogram, and no aspirin pre-treatment before PCI. However, overall the model predicted bailout poorly (c-index = 0.64). The need for bailout cannot be reliably predicted using baseline characteristics. Patients experiencing complications have poor clinical outcomes despite bailout use of GP IIb/IIIa inhibitors.