To be of greatest use for the measurement of significant variations in treatment outcomes, a health status index should have ratio properties. Previous attempts to construct such a scale using the magnitude estimation technique have failed, principally because of an inappropriate choice for the zero point. By rejecting death as the logical zero point for a scale of social preference and choosing instead the absence of that being measured (dysfunction and discomfort), the authors attained reliable values from patients, for an illness index, using magnitude estimation. This type of scale allows comparative judgments between treatment methods to be made with greater validity. It should aid in the wise allocation of health care resources