To define the process of outpatient consultation, the authors conducted a prospective study of 716 consecutive outpatient consultations in a university-based primary care internal medicine practice. The overall consultation rate was 11.9 per 100 patient visits, with 78% of the referrals to other physicians and 22% to non-physician specialists. Consultation rates and patterns of referral varied little between physicians with different levels of experience. Eighteen per cent of the consultations resulted in a no-show by the patient to the consultant. Referring physicians received communications from the consultants 80.5% of the time when appointments were kept. By multivariate regression two variables were shown to be most important in determining the internist's overall satisfaction: 1) how well the consultant aided the internist in his ongoing management of the patient's problem, and 2) how well specific questions were addressed by the specialist. Other statistically significant variables were the clarity and promptness of the consultant's reply, the educational value of the consultation, and specific management recommendations made by the consultant. To improve the consultation process no-shows must be minimized, communication from the consultant maximized, and the interaction between the internist and the consultant bolstered.