Costs and prices per patient admission vary greatly across patients and across hospitals. The variance across hospitals is due in part to institutional differences--hospital size, teaching status, local labor costs, and so on--but also to differences in different hospitals' patients' conditions. For purposes of reimbursement under the Medicare program, differences in patients' conditions are typically accounted for by nothing more than noting their different DRGs, without regard to intra-DRG patient differences. My results, which are based on 3.6 million individual patients, show that differences in individual patient DRGs and in observable indicators of their conditions are more than just relevant or important: They are nearly dispositive, explaining over 80% of the total variation in net price per patient admission. Moreover, cross-DRG differences explain only half of that price variation, with the other half explained by patient-level intra-DRG differences in patient condition that are not accounted for by patients' DRG classifications. Further analysis shows that those hospitals that tend to attract the more complex DRGs tend also to attract those patients who are more expensive to treat within each of their DRGs.