Clinical trials in palliative care address mostly physical and psychosocial distress in patients and families. Because of the multidimensional aspects of most physical and psychosocial symptoms as well as the lack of generally accepted gold standards for many of these outcomes it is common to measure multiple outcomes. Crossover trials offer unique opportunities to address the main outcome of a clinical trial within the context of all other accompanying symptoms. However, they have the disadvantage of complex analysis, a changing baseline set of conditions, and the absence of a ‘‘responder’’ status. Parallel studies require larger sample sizes but they are usually safer to conduct and interpret. Whenever possible, randomized controlled trials should be preceded by pilot studies addressing the feasibility and the sensitivity of different outcomes to change in the time required. These pilot studies allow for appropriate modification in the final design of the randomized controlled trial. The global of expression of satisfaction with treatment is very important in palliative care and is frequently omitted. Until better information is available regarding ‘‘responder’’ status clinical trials should always report the median change in the symptom status so as to allow the reader to interpret the impact of the intervention on different outcomes. Questions and comments will be welcome.