OBJECTIVES: As utility-based health measures, the EQ-5D and HUI3 draw from a different theoretical basis than psychometrically-derived measures, yet both weighted utilities and unweighted approaches find common application as outcomes in clinical trials where inferential statistics are applied. The objective of this study was to compare the statistical properties of the EQ-5D and HUI3 health state classifier systems using weighted utility scores (WUS)to using unweighted simple summary scores (SSS). METHODS: WUS and SSS were calculated for EQ-5D and Health Utilities Index Mark 3 (HUI3) health state classifier systems using two longitudinal datasets (n(stroke)=124; n(rehabilitation)=264). Validity was evaluated using F-statistics from ANOVA-based known groups comparisons. Pre/post paired t-tests were used to evaluate responsiveness. The relative efficiency (RE) ratio of for each of the statistics was calculated between WUS and SSS, with differences outside the range of 0.8 to 1.25 (e.g. >25%) interpreted as substantial. RESULTS: Greater statistical power was associated with WUS groups than SSS in known groups comparisons for EQ-5D (50% of RE>1; no substantial differences) and HUI3 (100% of RE>1; 75% substantial). WUS tended to be more responsive for the HUI3 (62.5% of RE>1; 37.5% substantial), while SSS tended to be more responsive for the EQ-5D (87.5% of RE<1; 75% substantial). CONCLUSION: Inferential power based on relative efficiency in this study implied that the approach to scoring (WUS or SSS) can provide greater statistical efficiency, which can reduce sample size requirements and thus study costs, but neither approach was consistently advantageous in this regard. Functional form, differences in utility derivation, and complexity of the health classifier may explain, in part, RE differences between the EQ-5D and HUI3.