Although research in primary care has a higher profile than ever before, its impact on professional practice and on government planning often seems less than it should be. In the first part of the paper, the different research agendas of governments, health departments, professional associations and colleges, and of universities are explored. In the second part of the paper a research project which attempts to define and measure quality of care given to patients with a 'marker' health problem (arthritic pain) is developed from the stage of asking questions to interpreting findings. In the third part of the paper, a number of conflicts between research agendas, styles of research, and needs and expectations of different 'purchasers' and 'providers' are explored using the themes and the details of the earlier parts of the paper as illustration, and a model is constructed to help explain why research, practice and policy making often live less easily together than is good for each. The importance of creating a supportive climate for research, of providing adequate infrastructure, and of making appropriate training available is emphasized.