While not uncontested, the finding that liberal democracies rarely, if ever, fight wars against each other represents one of the seminal discoveries of international relations (IR) scholarship. Nevertheless, 'democratic peace theory' (DPT) – the body of scholarship that seeks to explain the democratic peace finding – still lacks a satisfactory explanation for this phenomenon. In this thesis, I argue that a primary source of this failure has been DPT's failure to recognize the importance of collective identification and trust for the eventuation of the 'liberal peace'. Building on existing DPT scholarship, most of it Realist or Rationalist in its inspiration, but also employing insights from Constructivist and Cognitivist scholarship, I develop a new model of how specific forms of collective identification can produce specific forms of trust. On this basis, I elaborate a new explanation of the liberal peace which sees it as arising out of a network of trusting liberal security communities. I then elaborate a new research design that enables a more rigorous and replicable empirical investigation of these ideas through the analysis of three historical cases studies, namely the Canada-USA, India-Pakistan, and France-Germany relationships. The results of this analysis support the plausibility of my theoretical framework, and also illuminate four additional findings. Specifically, I find that (1) IR scholarship needs a more nuanced understanding of the interaction between agents and structures; (2) 'institutionalized collaboration' is especially important for promoting collective identification; (3) DPT scholarship needs to focus more attention on the content of the narratives around which collective identification takes place; and (4) dramatic events play an important role in collective identification by triggering what I term catharses and epiphanies. I close the thesis by reviewing the implications of my findings for IR and for policymakers and by suggesting some areas worthy of additional research.