The randomised controlled trial is regarded as the most robust method of assessing the effectiveness of treatments, although it is also recognised as being a logistically difficult procedure in the field of palliative care. The imPaCT study was a randomised controlled trial of two levels of inpatient palliative care team intervention in a large teaching hospital in southwest England. 261 patients were recruited to the study between 1997 and 2000. A variety of standardised outcomes were used in the four-week follow-up time. This paper reflects on the issues that emerged in the collection, analysis and interpretation of the research findings. In particular three issues will be put forward for discussion:

1. The tension apparent in evaluating the process of delivering palliative care as opposed to the outcome of palliative care
   
2. The question of whether evaluating some of the parts of the service adds up to evaluating the whole service
   
3. The sensitivity of data capture instruments and the status of non-quantifiable data
   

Underlying these three issues is the question of how "effectiveness" is defined. It is suggested that definitional problems lie at the heart of the research agenda in palliative care, not just in terms of the palliative care constituency and remit, but also in terms of the basic expectations that are held of the difference or impact that palliative care intervention should bring.