OBJECTIVE: The publication of information about the performance of health-care providers is regarded as central to promoting greater accountability and empowering patients to exercise choice. The evidence suggests that the public is not very interested in accessing or using current sources of information. This study aimed to explore the information needs of patients in the context of UK primary care and to develop an information source about general practice services, designed to be usable by and useful to patients. DESIGN: An action research study making use of data from formal and informal interviews, focus groups, participant observation and document review. SETTING: The geographical areas covered by two Primary Care Trusts in the north of England and two Local Health Boards in south Wales. PARTICIPANTS: A partnership between 103 members of the public, general practice staff from 19 practices, NHS managers from four Primary Care Organizations and the research team. RESULTS: The public would like to know more about the quality and range of general practice services but current sources of information do not meet their needs. The public do not like league tables comparing the performance of practices and only a small number of people want to use comparative information to choose between practices. They seem to be more interested in the context and availability of services and the willingness of practices to improve, than in the practice's absolute or relative performance. They want to be clear about the source of the information so that they can make personal judgements about its veracity. Information is most likely to be useful if it adheres to the basic principles of cognitive science in terms of its structure, content and presentation format. Using these findings, paper and electronic prototype versions of a guide to general practice services have been developed. CONCLUSIONS: In order to maximize the potential use of performance information by the public it is necessary to move beyond provider-led and professionally constructed approaches to information provision and ensure that the public is actively involved in the development of information sources. Such involvement produces a different kind of information to that currently available to the public. The findings of this study have important implications for policy. Most importantly, it seems that the traditional consumerist model underlying a policy of making comparative performance information available to the public to enable them to exercise choice between primary care providers may not be appropriate. An alternative model of information provision, which recognizes the public's commitment to their practice and is integrated with 'soft' sources of knowledge is more likely to engage and be of use to the public.