Our objective was to compare three methods of risk adjustment used in estimates of medical care expenses. We used two-part models and 2001-2004 MEPS data to estimate total medical expenditures, expenditures on office-based physician services, and prescription drugs, for persons by whom they choose to be their usual source of care. Estimates were risk-adjusted using demographic covariates, health status variables and a diagnoses-based method (DxCG). We assessed the relative effect of these three adjustment methods in accurately estimating medical care expenses. We found the methods together accounted for 19% – 31% of the expenditure estimates. The diagnoses-based method is more effective when estimating expenditures on office-based physician services, and less effective when estimating expenditures for persons with sub-specialty physicians as their usual source of care.