The hospital-based generalist physician is a recent organizational innovation in health care in the United States. Does the hospitalist model provide improved health care? The answer to this question lies in a rigorous evaluation of the hospitalist system in the clinical setting. This paper describes key outcomes that need to be assessed and methodologic issues that need to be addressed when conducting and interpreting the results of evaluations of the hospitalist model. To provide evidence about the value of the hospitalist model, quality of care should be evaluated through the measurement of both processes and outcomes. The clinical processes assessed may include diagnostic tests and treatments that are causally linked with outcomes. A variety of outcomes can be measured, such as death, clinical or physiologic status, physical function, and psychological well-being. The analysis of resource use data can aid in an assessment of the relative costs and effectiveness of clinical services. An evaluation of the hospitalist model requires an adequate research design, which should include a precise definition of the model being studied, the selection of an appropriate comparison group, the collection of clinical and demographic information on patients, the specification of both process and outcome measures, and the use of statistical techniques that are appropriate to the questions being asked and the data being collected. The design must distinguish between outcomes attributable to the introduction of hospitalists and those attributable to other changes in medical treatments and the organization of care.