OBJECTIVES: :  Different multi-attribute utility (MAU) instruments are known to produce different values for “utility” and measure different constructs, despite the common label “utility”.  To date, the Multi-Instrument Comparison (MIC) project has been the largest comparative study of health and well-being instruments undertaken worldwide.  Here we report the first results from the German branch of the study. METHODS: :  1,269 German respondents (either healthy or suffering from defined chronic disorders, i.e., asthma, arthritis, cancer, depression, diabetes, hearing loss, heart disease) were recruited and participated in the study, completing various MAU instruments, including the EQ-5D, SF-6D, HUI3, 15D, QWB, AQoL(-4D and-8D).  Cross-validation tests drew heavily on correlation.  Preliminary findings, based upon Pearson correlation coefficients (indicating the extent to which changes in one variable correspond with changes in another), showed low correlations between measures of utility and measures of subjective well-being.  While preferences might differ from subjective well-being, their correlation might be higher.  Hence, a better measure should be intraclass correlation (ICC). RESULTS: :  Intraclass correlations between MAU instruments ranged from to 0.8 (HUI3 vs. AQoL-8D) to 0.4 (AQoL-4D vs. 15D).  Linear regression results, reflecting the comparative performance of the various MAU instruments with regard to changes in measured utilities (as applied in standard cost utility analysis), and detailed results including pairwise comparisons of instruments, especially as to sensitivity to changes in a given dimension, will be presented. CONCLUSIONS: :  A major conclusion of the present study is that, despite some similarity in the mean scores, the instruments tested are dissimilar with regard to virtually all other criteria used to compare them.  In effect, each instrument appears to measure a different construct of “health”.  Implications for the presumably “generic” measurement of “utility” may be far reaching and will be discussed.