Research projects in order to evaluate social and health care programmes are unusual in Sweden. Evaluation studies have had little or no discernible effect on the planning of services for the elderly which is largely the product of convention rather than rational thinking. It is difficult to carry out evaluation studies in a field that is rapidly changing. The variations in elderly care between different areas, between institutions and over time makes it difficult to devise general methods of evaluation. Of five recent and current Swedish studies two are treating innovations inside institutions, two deal with innovations outside institutions and one is a relocation study. One thing common to these studies is that they are administered by university departments. However, evaluation is too important to be a matter just for university disciplines. We should also encourage local initiatives to start more evaluation of 'their' interventions. It will then be a task for more centrally placed researchers to educate and guide the administrators and representatives from other areas in what respects the result of a local study can be generalized.