2183 among facilities have to invest more time and money utilizing health care than patients who can receive all their care in one place. Furthermore, an intervention structure may not be f lexible enough to accommodate evolving health goals. Whereas vertical structures may be appropriate in the short term, they are likely to become increasingly inappropriate in the longer term, as a population’s health care needs and demands change. The narrow scope and rigidity of vertical delivery is especially problematic when needs and demands are changing rapidly — for example, in countries undergoing rapid socioeconomic development with changing lifestyles, health risk taking, and care-seeking behavior. In some situations, vertical interventions may draw resources such as health care workers away from general health systems, weakening general health care delivery.5 Furthermore, a predominantly disease-specific focus may distract from long-term planning priorities, such as training a generalist health care workforce. However, it is also possible that vertical programs can provide the motivation and resources to build specific types of capacity that can benefit the entire health system, as the PEPFAR-supported DELIVER Project of the U.S. Agency for International Development demonstrates with respect to drug supply chains. Finally, an intervention structure may impede evaluation. Donor organizations increasingly require evaluation of interventions’ impact on population health. Horizontal interventions such as building infrastructure, training health care workers, or integrating disease-specific programs into general health care systems are often difficult to evaluate, because their effects are realized over the medium and long term — and because they affect multiple disease outcomes and are, on their own, insufficient to guarantee that effective health care is delivered. The simultaneous trends toward horizontal interventions and better evaluation of impact thus run counter to one another. Donors and governments need to carefully consider the increased difficulty of evaluating impact when they shift from vertical to more horizontal delivery structures. New methods for evaluating horizontal interventions are also needed. The apparent policy shift in global health funding from vertical toward horizontal interventions carries the promise of improving health care delivery in developing countries — but also the danger of failure, if changes in delivery structures are undertaken without regard for local health threats, context, and capacities. Global health organizations must work with policymakers in countries receiving funding to identify the optimal mix of intervention structures for each context. Their deliberations will need to address the effect of context on the feasibility, efficiency, flexibility, and ease of evaluation of various intervention structures. Disclosure forms provided by the authors are available with the full text of this article at NEJM.org.