This paper discusses how Middle Level Theory is becoming an important tool in the evaluation of international development programmes by connecting abstract theory to project level theory. The paper discusses three defining features of how mid-level theorising generates transferable knowledge across disciplines and settings; it consolidates empirical regularities in human behaviour, it explains the diversity of outcomes observed across contexts, and it is explicit about causal principles observed in different settings. These attributes are illustrated using examples from interventions in international development that show the potential of the method in improving the design of international development programmes.