Two methodological concerns in utility estimation, the development of health state descriptions (scenarios) and the interpretation of interval scale anchor points, are examined in the context of disease-specific cost utility analyses (CUA). It is contended that results in CUA can be fundamentally biased by: (i) how the information presented in a scenario is generated; and (ii) the researcher's 'definition' of anchor points, when these are used as bounds to the interval scale. A number of recommendations are made, in particular for a more explicit reporting of these issues in CUA, to facilitate greater consistency in the application of utility measurement techniques.