Small sample size represents a vexing problem in evaluating medical management programs. Physician practices may have panel sizes of only 1000 total patients, disease management programs may enroll only several hundred patients, and hospitals may see only several dozen of a particular type of case per year. Given the high variation that occurs in the presentation, severity, and cost of medical conditions, estimation of either quality impact or economic return on investment can be difficult or impossible. This paper seeks to describe common approaches to this issue that are in practice within the industry today and to discuss limitations, alternative strategies, and consequences of method choices. A number of strategies exist to deal with this situation, but ongoing recognition of this issue in the planning phases of initiatives remains problematic. Significant linkage of financial rewards or penalties to poorly evaluated programs carries significant liabilities for the unsophisticated developers of such systems.