OBJECTIVES: The Blinder-Oaxaca (BO) decomposition helps to identify the causes of disparity between two groups based on observed characteristics, and quantifies the residual disparity due to unobserved heterogeneity. This study uses the non-linear BO decomposition, an extension of the original BO decomposition, to identify and quantify the observed causes behind the racial/ethnic disparity in anti-obesity medication use among US adults who met the medical criteria to receive these medications. METHODS: The Medical Expenditure Panel Survey data, 2002-2007, was used in this retrospective cross-sectional study. The medical criteria was defined as adults (age>=20 years) with body mass index (BMI)>30, or with BMI≥27 and comorbidities such as hypertension, cardiovascular diseases, diabetes or hyperlipidemia. The non-linear BO decomposition method was used to identify the contribution of each observed socio-demographic and health characteristic in explaining the racial/ethnic disparity and to estimate the residual unexplained disparity. RESULTS: A total of 395,319,750 adults met medical criteria from 2002-2007; of which, 0.61% used anti-obesity medications. The proportion of medically eligible Blacks using anti-obesity medication was 33% lower than Whites. The non-linear BO decomposition method revealed that only marital status explained the racial/ethnic disparity among the observed characteristics. Blacks were younger, had a higher proportion of women in the sample, and had a higher BMI, and these characteristics were the chief determinants of increased anti-obesity medication use. Thus, the decomposition estimated that if the blacks had similar characteristics as the whites the observed disparity in the sample would have been 58% higher.  CONCLUSIONS: The decomposition revealed interesting but disconcerting findings. None of the differences in the observed socio-demographic and health characteristics (except marital status) successfully explained the racial disparity. Thus, further exploratory studies examining racial differences in individual beliefs, behavioral patterns and provider prescription patterns are vital to understand these disparities.