OBJECTIVE: To analyze differences in the determination of healthcare expenditures among racial/ethnic groups. DATA SOURCE: The 1987 National Medical Expenditure Survey, a nationally representative sample of the United States population. (Nomenclature reflects racial/ethnic categories as used in the Survey.) STUDY DESIGN: The design was to estimate completely separate demand systems for blacks, Hispanics, and whites, perform statistical tests for the appropriateness of such separation, and carry out various simulations of healthcare expenditures. DATA COLLECTION/EXTRACTION METHODS: All black, white, and Hispanic persons in the 1987 NMES Household Survey were used in this analysis. PRINCIPAL FINDINGS: Several of the differences among the equations for the three racial/ethnic groups appear to be related to access to care, particularly between Hispanics and whites, and to a lesser degree between blacks and whites. Simulations indicated that most of the differences in healthcare spending were due on net to differences in characteristics of the sampled persons and their environments. However, for Hispanics relative to both blacks and whites, some of the differences in total expenditures were also due to differences in the behavior embodied in the equations. CONCLUSIONS: It would be inadequate, and possibly misleading, to allow for differences in health expenditures by simply including dummy variables for blacks, Hispanics, and/or whites in pooled equations estimated for the entire sample. Studies that allow one to analyze the institutional and behavioral aspects of healthcare spending in greater detail are needed for a better understanding of these racial/ethnic differences.