The availability of different morbidity systems derived from the same raw clinical data raises a number of questions. How much do clinical measures of morbidity improve the performance of mod-els of patient cost compared with simpler models based on age and gender? Do measures which account for multi morbidity per-form better than simpler morbidity systems which do not allow for possible interactions amongst diagnoses? Does the morbidity system used affect the relative performance of different estimators?In terms of the two policy motives for estimating cost models: how do capitation payments based on detailed clinical measures differ from those based only on age and sex? Are estimates of the relationship between socioeconomic status and cost sensitive to which morbidity system is used?