OBJECTIVES: Evidence on the use of newer antiplatelet agents and their cost implications remains scarce.  Previous research has shown a shorter average hospital length of stay for prasugrel-treated patients compared to clopidogrel-treated patients. We analyzed a large geographically diverse database from the US and compared cost of hospitalization for patients with acute coronary syndrome (ACS) who have undergone percutaneous coronary intervention (PCI) and who received either clopidogrel or prasugrel. METHODS: Using a large representative US database maintained by PREMIER, we analyzed patient characteristics and total hospitalization costs during the index (first) hospitalization among ACS-PCI patients treated with clopidogrel or prasugrel between July 2009 and June 2011. Analysis included patients treated with prasugrel who were on-label and clopidogrel-treated patients who would have been eligible for prasugrel treatment per the label.  Observed costs were analyzed unadjusted and adjusted for baseline differences using a generalized linear model with a gamma distribution and log link function with propensity score stratification. RESULTS: Data were available for 75,315 patients who received clopidogrel and 9,483 patients who received prasugrel during their hospitalization. The observed mean hospitalization costs (SD) for clopidogrel and prasugrel, respectively, were $17,519 ($2,548) and $17,136 ($2,562). Mean costs for clopidogrel and prasugrel recipients, respectively, were $16,937 ($2,162) and $16,664 ($2,137) for STEMI, $17,926 ($2,747) and $17,511 ($2,849) for NSTEMI, and $17,900 ($2,665) and $17,393 ($2,676) for UA (all comparisons, P<0.001). The adjusted results showed prasugrel-treated patients cost as much as $882 less than clopidogrel-treated patients (P<0.001) during the index hospital stay. CONCLUSIONS: Prasugrel-treated patients used fewer health care resources compared to clopidogrel-treated patients during the index hospital stay, as measured by hospital costs. Similar results were obtained after adjusting for patient demographics and clinical characteristics. The potential for unmeasured confounder bias is a limitation in this real-world observational research.