Background— Practice patterns in anticoagulant strategies used during percutaneous coronary intervention (PCI) in the United States for patients with non–ST-segment–elevation myocardial infarction and the comparative outcomes between bivalirudin and unfractionated heparin (UFH) have not been well described. Methods and Results— Trends in anticoagulant use were examined among 553 562 PCIs performed by 9254 operators at 1538 hospitals for non–ST-segment–elevation myocardial infarction from 2009 to 2014 within the CathPCI Registry. To compare bivalirudin with UFH, propensity score matching and instrumental variable (IV) methods with operator preference for bivalirudin as the instrument were used. To determine whether differences in outcomes were because of differences in glycoprotein IIb/IIIa inhibitor (GPI) use, a test of mediation was performed using the IV. Outcomes were in-hospital bleeding and mortality. Bivalirudin use increased from 2009 to 2013 but declined during 2014. GPI use was 50.5% during UFH PCIs and 12.0% during bivalirudin PCIs. Before GPI adjustment, bleeding reductions with bivalirudin ranged from 2.04% (IV: 95% confidence interval [CI]: 1.81%, 2.27%) to 2.29% (propensity score: 95% CI: 2.14%, 2.44%) and mortality reductions ranged from 0.16% (IV: 95% CI: 0.03%, 0.28%) to 0.25% (propensity score: 95% CI: 0.17%, 0.33%). After GPI adjustment in the IV, more than half the bleeding reduction with bivalirudin was because of the lower use of GPIs (risk difference, −0.84%; 95% CI: −1.11%, −0.57%), and no survival benefit was apparent (risk difference, −0.10%; 95% CI: −0.24%, 0.05%). Bleeding reductions with bivalirudin were largest for transfemoral PCI (GPI-adjusted risk difference, −1.11%; 95% CI: −1.43%, −0.80%) and negligible for transradial PCI (GPI-adjusted risk difference, 0.09%; 95% CI: −0.32%, 0.50%). Conclusions— In the largest comparative analysis of bivalirudin versus UFH for non–ST-segment–elevation myocardial infarction to date, bivalirudin was associated with lower in-hospital bleeding and mortality given current practices with respect to GPI use and access site. Bleeding differences were, in part, explained by the greater use of GPIs with UFH. Reductions in bleeding were largest among those undergoing transfemoral PCI, whereas no bleeding benefit was observed for those treated with transradial PCI.