OBJECTIVES: To evaluate the measurement properties of the EQ-5D profile augmented with its 0-100 mm visual analogue scale (VAS) and a 5-point summary self-rating of health (SRH). METHODS: We used data from 4,001 adults from the 2003 MEPS who had ≥1 of 7 most prevalent chronic conditions and completed the EQ-5D, VAS, and SRH. The original 101 VAS categories were collapsed into a 9-category item with sufficient responses in each category. Five SRH categories included “excellent”, “very good”, “good”, “fair” and “poor”. The Rasch rating scale and partial credit models were used to calibrate the EQ-5D and the single items, respectively. Calibrations were conducted using 4 different combinations: EQ-5D alone; EQ-5D+SRH; EQ-5D+VAS, and EQ-5D+SRH+VAS. Model goodness-of-fit was assessed in each disease group using INFIT mean squares (≤1.40). Principal Component Analysis of Rasch Residuals was used to confirm dimensionality examining the proportion of total variance explained by Rasch scale, person measures and item measures, respectively. RESULTS: Respondents were predominantly white, female, middle aged and suffered most commonly from hypertension (32%), diabetes (17%) and depression (15%). EQ-5D item “anxiety/depression” consistently showed misfit to the model across 7 conditions when EQ-5D was evaluated alone. The inclusion of VAS and/or SRH not only improved model fit, but also increased overall variances explained, and improved overall distribution of persons and items along the latent health trait. Specifically, when both items were included, 4 groups showed good model fit (mean squares ≤1.40). Consistently across all groups, VAS captured more person measures while SRH captured more item measures. CONCLUSIONS: The EQ-5D’s measurement qualities are enhanced by the inclusion of VAS/SRH, which captures integral aspect of self-valuations on health that are possibly overlooked by the EQ-5D. The EQ-5D+VAS+SRH may serve as a suitable measurement framework for deriving population preference-weights. Consequently, a new valuation algorithm is called for.