We have studied the knowledge of students and clinicians that they actually used to follow up four problems in general medicine. Some unexpected similarities and differences in their readily accessible or primary knowledge were found in groups from first-year clinical medical students to consultants. There is no linear increase in the quantity of primary knowledge with experience, but qualitative changes are very much more important. At all levels individuals show a remarkable dissimilarity in the knowledge they use to solve clinical problems, so that the bulk of personal knowledge used is individual. Comparatively rarely do people use the same knowledge to solve any one clinical problem. However, there is evidence of a general tendency towards increasing uniformity in knowledge as a result of the medical school years. After houseman level, individuality increases again. These changes result in consultants achieving an identical profile to first-year clinical medical students in terms of the extent to which their primary knowledge bases are held individually or in common. These findings suggest that consultant expertise is actually based on individual experience rather than a common core of knowledge. The influence of medical school in providing such a common knowledge base is (1) limited and (2) reversed by clinical practice. The results suggest the necessity for increased vocational and practice-oriented components in medical education, particularly in postgraduate education.