OBJECTIVES: To compare the performance of Health-related Quality of Life Comorbidity Index (HRQL-CI), D’Hoore Comorbidity Index, Elixhauser Comorbidity Measures, and Disease Count in risk-adjusting HRQL in diabetes patients.  METHODS: The 2008 Medical Expenditure Panel Survey (MEPS) including adults ≥18 years with diagnosis of diabetes (ICD9 = 250) was used to evaluate the performance of four comorbidity measures for risk-adjusting Short Form-12 Physical Component Score (PCS) and Mental Component Score (MCS). The comorbidity measures included  D’Hoore adaptation of Charlson comorbidity index (identified using 3 digit ICD9 codes), Elixhauser Comorbidity Measures (identified using 3 digit ICD9 codes), HRQL-CI (identified using Agency for Healthcare Research and Quality Clinical Classification Codes), and Disease Count (identified using ICD9 codes). Baseline model included age, gender and race. Weighted linear regression analyses were used to risk-adjust PCS and MCS. Model performance was evaluated using adjusted-R squares.  RESULTS: According to the MEPS, there were 20.9 million diabetic patients (unweighted sample 2155) in 2008 with mean age 59.4 (±14.2) years.  The mean (± SD) PCS and MCS scores for diabetes patients were 40.51 (±0.37) and 48.94 (±0.34), respectively.  The baseline model (age, gender and race) performance was: (PCS: R2=0.0920; MCS: 0.0207). The HRQL-CI explained most of the variation in the PCS scores (R2=0.3197). The disease count (R2=0.2704) and Elixhauser (R2=0.2679) explained similar amount of variation in PCS. Unlike PCS, Elixhauser (R2= 0.1865) performed best in explaining variance in MCS, followed by HRQL-CI (R2=0.1493) and Disease count (R2=0.0905). D’Hoore Index did not perform well in both PCS (R2=0.1829) and MCS (R2=0.0516).  CONCLUSIONS: Recently developed HRQL-CI performed better than other comorbidity measures for risk adjusting PCS; however, Elixhauser comorbidity measures performed better than HRQL-CI for MCS.