Health programmers and researchers must collaborate despite different mandates and technical languages. A results framework is a simple model that both disciplines can use to understand complexity, clarify assumptions and hypotheses, design programs, and ask questions to inform action research. Typically, a health program's results framework has 3 tiers and 6 boxes: a base of 4 health service intermediate results (access, quality, demand, and environment), which lead to a midlevel strategic objective (use of life-saving intervention), which leads to the goal (improved health). A situation analysis directly informs intervention selection; more difficult is selecting strategies to deliver the interventions, especially in settings of health system weakness. We propose menus and submenus of strategies to achieve each intermediate result, illustrate the use of the results framework in a program design and in clarifying research questions, and begin to propose a research agenda for "delivery scientists" responsible for recommending optimal investments to maximize use of interventions by those who need them most.