BACKGROUND: Inappropriate and costly GP prescribing is a major problem facing Primary Care Trusts. Educational outreach into practices, alongside other measures, such as audit and feedback, have the potential to enable GP prescribing to become more evidence based. High GP prescribing costs are associated with GPs who see drug company representatives; tend to end consultations with prescriptions; and 'try out' new drugs on an 'ad hoc basis' and use this as evidence of the drug's effect. An educational intervention called 'reflective practice' was developed to meet these and other educational needs. The design of the intervention was informed by studies that have identified the pre-requisites of successful behaviour change in general practice. OBJECTIVES: The study investigated the following: (i) Is it feasible for GPs to attend the sessions included in the educational intervention? (ii) Is the intervention acceptable to the participants and the session facilitators? (iii) What are the barriers to the group educational processes, and how can these be overcome? METHODS: Four practices were recruited in South West England, all of them experiencing problems with prescribing appropriateness and cost. Reflective practice sessions (including a video-taped scenario) were run in each of these practices and qualitative methods were used to explore the complex attitudes and behaviour of the participants. A researcher observed and audio-taped sessions in each practice. At the end of the programme, a sample of doctors and all the facilitators were interviewed about their experiences. The recorded data were transcribed and analysed using standard qualitative methods. RESULTS: The doctors in the largest partnerships were those who had the greatest difficulty in attending the sessions. Elsewhere, doctors were also reluctant to become involved because of previous experience of top-down managerial initiatives about prescribing quality. Facilitators came from a broad range of professional backgrounds. While knowledge of prescribed drug management issues was important, the professional background of the facilitator was less important than group facilitation skills in creating a group process which participating GPs found satisfactory. The video-taped scenario was found to be useful to set the scene for the discussion. Preserving the anonymity of responses of the GPs in the initial stages of the sessions was important in ensuring honesty in the discussion. Reaching a consensus on management of common conditions was sometimes difficult, partly because the use of the term 'best buy' implies economic pressures, rather than benefits to patients, and partly because of the value with which GPs regard the concept of clinical autonomy. 'Reflective Practice' appeared to have the potential to make GPs aware of the link to be made between their clinical management decisions and the evidence provided by the British National Formulary and Clinical Evidence. CONCLUSION: The study indicates the importance of preparing the practice adequately, including providing protected time for all GPs to attend the educational intervention. Scenarios and the structure of the sessions need to make more explicit the links between everyday practice and published evidence of effectiveness. Emphasis on cost-effectiveness may be counterproductive and wider benefits need to be emphasized. We have also identified the skill profile of the facilitator role. Our study indicates a need for a clearer understanding of GPs' perception of clinical autonomy and how this conflicts with the goal of agreement on practice guidelines for treatment. The intervention is now ripe for further development, perhaps by integrating it with other interventions to change professional behaviour. The improved intervention should then be evaluated in a randomized controlled trial.