Patient preference-based health-related quality of life measures (utilities) are a critical input in cost-effectiveness analyses of pharmaceuticals and other health care technologies. Over recent years it has become more common for utility data to be collected alongside key clinical data within the pivotal Phase III trials. However, utility data are often not available for all patients throughout the entire course of the trial. The authors discuss the concept that there are unique characteristics of utility data that need to be considered when dealing with missing values, including the large inter-patient variability typically present at baseline. Missing data arise because i) patients become more ill and are less able to complete patient-reported instruments, ii) patients die during the course of the trial, or iii) patients are censored at later stages of the trial due to rolling recruitment. Whilst this is often a problem in oncology trials, it is also a consideration in other interventional and observational research designed to inform pharmacoeconomic evaluations. Situations where individual patient data are accessible and where only summary statistics are available are discussed. The practical considerations of how such re-analysed data should then be included within an economic model are discussed, given the manner in which utilities are incorporated will vary depending on the nature of the health states used.