While logic models such as logframes and results chains represent programs and policies as linear and predictable, this session will explore how program theory and logic models can be used for systemic evaluations, addressing relationships, perspectives and boundaries, and recognising interdependency, multidimensionality, and interaction. Program theory can represent complicated aspects of interventions - such multiple organizations working in parallel or sequence; interventions that work at multiple levels (eg individual, site, organization); multiple causal strands within an intervention; and interventions that both influence and are influenced by the context - and complex aspects - where it is adaptive, responsive, emergent and unpredictable. It also requires thoughtful decisions about where to draw the boundaries of what is being represented in the program theory. Ultimately everything is connected, but a logic model cannot include everything. The session will also explore the barriers to using more systemic versions of program theory and how these might be overcome.