A randomized controlled trial was undertaken to evaluate the effectiveness of a new model for providing urban general practice attachments for final-year medical students at the Flinders University of South Australia. All the student groups in that year were randomized prospectively to either the standard student attachment, as run by the university, or to an attachment organized by a project team from a local network of general practitioners. Students in the intervention group had their personal learning goals assessed and matched with their general practice preceptors, and the students were set a task that developed their contact with other health resources in the community. Results from an evaluation questionnaire completed by the students at the end of their terms showed that the students in the intervention group rated their general practice preceptors more highly, had more contact with allied health and community organizations, felt that they had met their own learning goals to a greater extent, and enjoyed their term more. Student examination results showed that the students in the intervention group did not perform as well in one of the four areas of their end of term examination as did the students in the standard attachment. The additional cost of providing the intervention was estimated to be A$340 per student. We conclude that long-term decisions about adopting this new model of organizing general practice attachments on a wider scale will need to balance the apparent benefits against the increased resources required.