Background Most analyses of complications after percutaneous coronary intervention have been limited to angiographic predictors of abrupt closure. We sought to determine the relation between baseline clinical and angiographic characteristics and clinical ischemic events and whether treatment with the platelet glycoprotein IIb/IIIa receptor antagonist c7E3 reduced ischemic events differentially in patients with distinct lesion morphologic characteristics. In the EPIC trial, a bolus and infusion of c7E3 decreased the 30-day incidence of death, myocardial infarction, and need for revascularization by 35% in 2099 high-risk patients. Methods We used logistic regression modeling to determine the relations between these patients’ baseline clinical and angiographic characteristics and the composite primary end point. We also constructed multivariable models with interaction terms to assess treatment effect on prespecified, core laboratory–assessed, coronary morphologic characteristics. Results The most important predictors of a poor outcome were low weight (chi-square = 10.5, P =.001) and preprocedural percent stenosis (chi-square = 15.0, P <.001). History of hypertension, nonwhite race, and peripheral vascular disease were also associated with an increased risk, as were all measures of lesion complexity except calcification and presence of a side branch. The treatment benefit with abciximab was significantly greater with less complex than with more complex lesion morphologic characteristics. Conclusions Future risk models should include these baseline characteristics to define the risk for ischemic complications in individual patients, and treatment with abciximab should not be predicated on lesion morphologic findings alone. (Am Heart J 1999;137:264-73.)