Since the adoption of Medicare's Prospective Payment System (PPS), critics have raised concerns about the degree of variation in cost per case due to variability in severity of illness within DRGs. Using hospital (Part A) costs for all Medicare beneficiaries hospitalized in Michigan during 1982, this paper applies multivariate techniques to test the ability of two state-of-the-art case-mix measures to either replace or act as a severity modifier to DRGs. When compared with DRGs, neither Patient Management Categories (PMCs) nor Disease Staging explains any more of the variation in costs. Results show that DRGs explain 33%, while PMCs and Disease Staging explain 26% and 17%, respectively. The use of either alternative in tandem with DRGs explains only 1-2% more of the observed variation; however, both PMCs and Disease Staging identify subgroups of patients with significantly different costs. These findings suggest that the alternatives may prove modestly useful as DRG modifiers by identifying classes of patients that do not belong. More importantly, they offer a measure of "unrelated comorbidity" that may improve the performance of DRGs.