OBJECTIVE: There are standard data sources for estimating the costs of inpatient and outpatient physician procedures, outpatient visits, and drug costs for the US, but no standard data source for estimating the average cost per day in the hospital at various levels of care (i.e., intensive or critical care units (ICU/CCU), regular care). METHODS: A method for estimating the cost per day in the hospital for different levels of care and different diagnoses was developed using data from the Healthcare Cost and Utilization Project's (HCUP) 2002 Michigan State Inpatient Database (SID). Linear regression models without an intercept were used to estimate the relationship between total cost and length of stay at each level of care for selected diagnosis-related groups (DRGs). RESULTS: Nearly 1.3 millions inpatient stays were initially included in the Michigan SID. Across all records with complete data (n=787, 753), the incremental cost per day, in 2005 dollars, in the hospital, for all medical conditions, was $1237 (SE=$0.97). The incremental cost per day by level of care across all conditions was $2401 (SE=$3.60) in an ICU or CCU and $1122 (SE=$0.97) in regular care. The incremental cost per day in an ICU or CCU varied across DRGs analyzed, ranging from $1241 (SE=$18.27) for DRG 387 (n=1422), prematurity with major problems, to $3,315 (SE=$106.60) for DRG 105 (n=1220), cardiac valve and other major cardiothoracic procedures without cardiac catheterization. The incremental cost per day in regular care varied from $504.94 (SE=$4.67) for DRG 379 (n=1789), threatened abortion, to $3081 (SE=$119.23) for DRG 106 (n=164), coronary bypass with PTCA. CONCLUSIONS: A simple regression analysis method can be used with data from national or state inpatient datasets to estimate average cost per day in the hospital at different levels of care and for different diagnoses. These estimates can be used in cost-effectiveness studies.