To receive adequate training experience, resident panels in teaching clinics must have a sufficiently diverse patient case-mix. However, case-mix can differ from one resident panel to another, resulting in inconsistent training. Encounter data from primary care residency clinics at Massachusetts General Hospital from July 2008 to May 2010 (64 residents and ~3800 patients) were used to characterize patients by gender, age, major disease category (both acute and chronic, e.g., Cardio Acute, Cardio Chronic, etc., for a total of 44 disease categories), and number of disease categories. Imbalance across resident panels was characterized by the standard deviation for disease category, patient panel size, and annual visit frequency. To balance case-mix in resident panels, patient reassignment algorithms were proposed. First, patients were sorted by complexity; then patients were allocated sequentially to the panel with the least overall complexity. Patient reassignment across resident panels was considered under 3 scenarios: 1) within preceptor, 2) within a group of preceptors, and 3) across the entire practice annually. were compared with case-mix (pre-July 2012) and post-July 2012. Results. All 3 reassignment algorithms produced significant reductions in standard deviation of either number of disease categories or diagnoses across residents when compared with baseline (pre-July 2012) and actual July 2012 reassignment. Reassignment across the clinic and group provided the best and second best scenarios, respectively, although both came at the cost of initially reduced patient-preceptor continuity. Systematically reallocating patient panels in teaching clinics potentially can improve the consistency and breadth of the educational experience. The method in principle can be extended to any target of health care system reform where there is patient or clinician turnover.