The quality of heath care in the United States is suboptimal and needs to be improved as part of increasing the value of costly health care services. Achieving broad quality improvements will require reproducing local quality improvement intervention (QII) successes on a larger scale. Such replication has been difficult to come by, however, because we don't understand the "how" of quality improvement very well. The goal of this analysis was to ascertain the predominant themes and patterns likely to be associated with producing successful QIIs. Cases were compared according to each dimension of the framework. The general approach involved establishing the range (the maximally diverse exemplars), the central tendency (the modal example) and the distribution (the pattern of variation) within each dimension and subcategory. The author assessed cases first on a univariate basis and then on a multivariate basis by grouping them according to more and less successful cases, different domains of care, and different degrees of organizational integration. The strengths of this study include its comparative case study design as well as its unique investigator-based sampling strategy, which sought to maximize the observed variation across cases while achieving an equal balance of "more" and "less" successful cases. Future research endeavors should attempt to operationalize and validate the archetypes suggested by this study. Doing so will produce broadly generalizable and practical tools for explaining how quality improvement results are generated, and for strategizing for success when implementing interventions in different settings.