Generic health status measures classify patients into different health states. For example, the EQ-5D descriptive system developed by the EuroQol Group classifies patients into 243 health states. Empirical values for the health states are available for only a selection (mostly 12 to 45) of these health states. Several parametric relationships between the descriptive system and the known values can be formulated to estimate the values for the unrecorded health states. This paper describes several of these modeling exercises in a comprehensible way, using the EQ-5D as an illustration. It is shown that the estimation task does not depend on the meaning of the values, but does depend on the selection of the empirically valued health states and the assumptions about the relationship between these values and the descriptive system.