We summarize issues that arise when considering quality of life (QOL) data in cancer clinical trials, especially those related to missing data. We describe different types of missing data mechanisms, and discuss ways of assessing and testing missing data mechanisms. A section on presentation of study design and results describes how graphical displays can effectively document the extent of the missing data problem, as well as describe its impact on interpretation of results. Finally, we describe several different statistical methods used to analyse repeated measures, with an emphasis on their properties and their ability to adequately handle different types of missing data mechanisms. We make recommendations as to the most appropriate methods, and suggest important directions for future research.