The Institute of Medicine argues that poorly designed delivery systems are a major cause of low-quality care in the United States but does not present methods for evaluating whether its recommendations, when implemented by a health care organization, actually improve quality of care. We describe how time-series study designs using individual-level longitudinal data can be applied to address methodological challenges in our evaluation of the impact of the Group Health Cooperative "Access Initiative," an integrated set of 7 "patient-centered" reforms in its integrated delivery system that are consistent with the Institute of Medicine's recommendations. The methods may be generalizable to evaluating similar reforms in other integrated delivery systems with representative patient and physician data sources.