Palliative care research is often said to be particularly challenging since the study population consists of patients who are frail, both physically and psychologically. This raises the question of the clinical and ethical justifiability of conducting clinical trials among these patients. Nevertheless, in several areas, we know far too little about what might be optimal treatment and care for dying patients and since the provision of both is indeed a medical duty, this seems in itself a very strong reason for palliative care research. There is also the issue of respect for individual autonomy. It is problematic not to let patients decide for themselves whether or not they would like to participate in clinical trials or other research a question they would not even have the opportunity to say "yes" or "no" to if it were the case that patients receiving palliative care were seen as unfit for research. The other side of the coin would then be the problem of paternalism towards patients; or what might be termed "unintended paternalism" - i.e., the kind of paternalism that follows by accident (unintentionally) when health care workers are too eager in protecting their patients against various perceived harms. While such protection is, in itself, both desireable and admirable, it runs the risk of paternalistic action, even without intent. This paper starts with some clinical observations regarding the problem of empirical autonomy in the seriously ill - pointing to factors (such as depression and hopelessness) that would contribute to a reduced level of autonomy in patients. Next, a philosophical-ethical analysis of the central concepts "autonomy" and "paternalism" is undertaken in order to achieve a better understanding of these frequently invoked, but seldom explained, concepts and their relevance to palliative care research.