OBJECTIVES: To examine how obesity and quality of life (QOL) affect medical (MC) and lost productivity costs (LPC) in U.S. adults with diabetes.   METHODS: A cross sectional study design was applied using the 2003-2006 Medical Expenditure Panel Survey data. The study population consisted of patients aged 18-64 with diabetes, but without immunodeficiency, malignancy, kidney dialysis, or underweight. Diabetic patients’ QOL were classified as low (≤25%), medium (>25%-<75%), and high (≥75%) based on quartiles of the SF-12 physical component scale(PCS-12) and mental component scale(MCS-12) respectively.  Patients’ BMI was classified as normal, overweight, and obese. Diabetic complications included heart disease, stroke, retinopathy, neuropathy, and nephropathy.  MCs were estimated using a generalized linear regression model with log link and gamma distribution after adjusting for patient characteristics and comorbidities.   LPC were calculated based on missed working days due to illness and average hourly wage using a two part model for working adults.  All costs were converted to 2008 U.S. dollars.   RESULTS: The study identified 3,621 diabetic patients.  Patients with low PCS-12 had more complications, compared to those with high PCS-12 (17.0% vs. 4.4%). A similar trend was also observed for MCS-12 (11.8% vs. 8.2%).  In general, patients with lower PCS-12 had higher MC ($12,203 in low-QOL vs. $3,172 in high-QOL) and LPC ($1,632 in low-QOL vs. $293 in high-QOL).  A similar trend was found in the relationship between MCS-12 and both costs, but this relationship was weaker than that of PCS-12.   Among patients with medium to high PCS-12, obese patients had higher MC and LPC than normal or overweight patients.  CONCLUSIONS: Lower levels of QOL were associated with a higher economic burden on diabetic patients, especially the physical QOL component.  Among diabetic patients with higher levels of physical QOL that are less confounded by disease severity and complications, the impact of obesity on economic burden was observed.