War is an undeniable phenomenon in international relations. But why do states fight each other? The quintessential question of the causes of war has drawn prolific scholarly attention ever since Thucydides' dissection of the Peloponnesian War (431-404 BCE). Yet little consensus has been reached. What makes war causes so elusive? One reason, I contend, is that ideographic and nomothetic explanations of war have traditionally been considered in opposition. There is, however, nothing mutually exclusive about a dynamic process and specific combinations of causal variables. Explaining the onset of war requires answers to both why and how war emerges – conditions and process. In this thesis, I therefore take a systems approach and argue that war, as any social phenomenon, can be conceived of and analyzed as an outcome of an endogenous process of actions and reactions which shape and, in turn, are shaped by context. Whereas natural systems consist of detectable secrets, social systems also consist of undetectable mysteries. Specifically, I develop a dynamic, nonlinear pre-war bargaining model which allows us to simulate how two adversarial states of configurable economic and military size, regime type, and leadership bargain over a contentious issue and decide whether to use military force to resolve it. By integrating the international, domestic, and individual-level material secrets and immaterial mysteries into a coherent dynamic whole, the model embraces the inherent complexity of social systems and enables us to study how small variations in initial conditions, changes produced by the bargaining process, and exogenous factors affect the stability of peace. The model thus brings together the process-oriented ideographic approach and counterfactual analysis, on the one hand, and the variable-oriented nomothetic approach on the other. Another reason for the elusiveness of war causes, I argue, is the paradigmatic divides and lack of scientific pluralism in IR. Since etiology is inextricably linked to methodology, these divides and the putative distinction between "causal" explanation and "interpretive" understanding have hampered scientific progress on the causes of war. Therefore, the model is constructed based on scientific pluralism and analytical eclecticism and builds on both Rationalism and Constructivism in its treatment of the first, second, and third images and their interdependence. The result is a highly configurable dynamic model which is intended to serve as a 'theoretical laboratory'. As an initial 'inference to the best explanation' to be improved in future iterations, this theoretical laboratory allows for theoretical process tracing and the identification of both agentic and structural necessary conditions counterfactuals. Etiologically, this amounts to a dynamic conjunctional concept of causes and causation in the form of INUS conditions and complexes. By simulating hundreds of thousands of perfectly replicable cases with minor variations over militarily symmetric and asymmetric dyads with states of different regime types, economic properties, and rational and sub-rational leaders, the model enables sensitivity analysis and both case-specific and general explanations of war onset.