OBJECTIVES: To compare the difference in the relationship between the BMI and health care expenditures in subjects with and without cardiovascular disease (CVD) using Medical Expenditure Panel Survey (MEPS) METHODS: We performed cross-sectional analyses of 11,385 adults from the 2005 MEPS, a nationally representative survey of noninstitutionalized civilian population in U.S. Subjects with CVD (coronary heart disease, myocardial infarction, stroke, and hypertension) were determined from self-reports. Mean expenditures per capita were estimated for NIH BMI categories (under, normal, overweight, obese I, II, and III) using a two-part exponential conditional model (ECM) adjusted for age, race, wage, occupation, type of health insurance, degree level, and smoking status. The first part of the model was logistic regression to predict the probability of incurring any expenditures. For the second part, we used ECM since the log-scale expenditure data was not leptokurtotic and was heteroscedastic. We performed Box-Cox test and Park test to find the appropriate link function and distribution family. Average expenditures in 2005 U.S. dollars were calculated by multiplying each person's probability of incurring any expense and predicted expenditures. RESULTS: About 67% of subjects with CVD (N=2596) and 65% of subjects without CVD (N=8789) were overweight or obese. Using gamma distribution with log link function, mean expenditures in CVD-group by BMI categories were $3247, $3040, $3098, $2966, $3500, and $3375 (p=non-significant). Those in subjects without CVD were $1857, $2327, $2389, $2534, $3179, and $3783 (p<0.001). Age, smoking status, and Medicare were significantly associated with expenditures in CVD-group. CONCLUSIONS: Health care-expenditures did not significantly differ among BMI categories in subjects with CVD whereas health care-expenditures were increasing with BMI in subjects without CVD. This could be due to the influence of CVD-care costs across BMI categories in CVD-group. As opposed to the previous finding of 'obesity paradox' - a paradoxical decrease in morbidity and mortality with increasing BMI in subjects with CVD, we did not find a significant 'obesity paradox' in health care-expenditures in subjects with CVD.