Survival and quality of life are the two end-points of cancer treatment. Recommendations are available as to how quality of life should be assessed in clinical studies in order for quantitative scores to be provided for it, along with survival data. The two measurements can be combined, and their total should yield a definition of the state of the art in cancer treatment with respect to both the quantity and the quality of life. However, at the individual level, quality of life can be traded off against survival only by taking into account the patient's attitude toward risk. More importantly, the quality of health status may be valued in a completely different way from patient to patient. Thus, when a clinical choice is to be made from among different treatment options, the ethical principle of respect for the patient's autonomy would require that the patient be informed of their possible respective outcomes, and allowed to provide his/her own assessment of the quality of life associated with these outcomes. This might have consequences for quality of life assessment in clinical studies, which should not only quantify average scores for treatment comparisons, but also provide a 'health state description' with respect to the aftermath of each of the treatment options. These data could be formalized in the already proposed decision instruments incorporating clinical scenarios for patient information. In any case, this should allow individualized clinical decisions incorporating each patient's preferences for the quality of his/her expected life span. Providing average quality of life scores may be useful, indeed, for population-based health decisions, as are those on resource allocation and those on registration of new drugs.