OBJECTIVES: The Model of Health Services Use is an analytic framework incorporating predisposing, enabling, and need predictor variables to explain patients' use of health care resources. We applied this model to the consolidated year 2000 Medical Expenditure Panel Survey (MEPS), a representative survey of the US civilian, non-institutionalized population, to compare the influence of various need variables, perceived health status, on the outcome of total health-related expenditures. METHODS: Multivariate linear regression models were developed, maintaining a core set of predictor (predisposing and enabling) and dependent (total health care expenditures) variables, varying only the need predictor variables. Predisposing variables include age, gender, race, education, and marital status; enabling variables include insurance type, employment status, family size, and household income; need variables included summary scores of two general health status measures (SF-12 PCS and MCS and the EQ-5D –Index and VAS), and single-item core MEPS questions of perceived mental and physical health status. Analysis took into account the complex design of the dataset. The R2 of each model is presented for descriptive comparison. RESULTS: Data from this MEPS dataset was obtained from 16076 respondents, representing over 209 million US residents. The base model, containing only predisposing and enabling variables, had an R2 of 0.064. The models using the single-item core questions of perceived mental and physical health status separately yielded R2 values of 0.073 and 0.104, respectively, with an R2 of 0.105 when both were included. R2 values for the models containing the SF-12 MCS, SF-12 PCS, EQ-5D index, and EQ-5D VAS individually were 0.068, 0.084, 0.065, and 0.071, respectively. Including the SF-12 MCS and PCS together in one model, the R2 was 0.086. CONCLUSION: The results imply that the single-item core health-status questions used by the MEPS perform marginally better than the SF-12 or EQ-5D to explain total expenditures.