In the present drive towards evidence-based health care it is essential for nurses to be able to define and measure their contribution in the health services and thereby ensure that resources are deployed to provide maximum improvement in the health of the population. Yet there exist few rigorous studies which examine the impact of the nursing contribution on patient health gain. One explanation for this is the sheer complexity of outcome measurement. This paper considers the multitude of factors which must be taken into account when designing studies to measure the impact of interventions for people with serious mental health problems. The heterogeneous nature of the population, the range of services these people might use, the composite of possible interventions and the paucity of adequate measurement tools are among the issues to be tackled. Although no single methodology can be prescribed, a number of principles are offered to guide study design.