This dissertation examines the strategic choices of multinational enterprises (MNEs) in host countries that become engaged in war. By combining the resource-based view and resource management theory, and drawing additional insights from research on real options and foreign strategic exit, I link the costs attributable to war to the strategic responses of the MNE at the subsidiary level in a novel firm-vulnerability framework. In particular, I develop theory regarding whether a subsidiary will exit from a host country, and if so, the timing (early or late) and mode (whole or partial) of exit. I test my hypotheses on a sample of 626 subsidiaries from 386 Japanese MNEs representing 51 industries in 23 countries at war, both interstate and civil, over the period 1988 to 2006. In analyzing the exit likelihood and timing decisions with time-varying covariates, I employ an extended Cox proportional hazard model, which allows for random-effects modeling of predictor variables at the subsidiary, parent MNE, and host country levels. To determine the exit mode of subsidiaries that choose exit over staying, I use binomial logit models. To correct for potential sample selection bias, I replicate my exit mode results with a Heckman probit model. My findings suggest that iv increasing strategic flexibility can counterbalance the potential disadvantages associated with leveraging strategically salient resources in high-risk locations. In examining war as a broad-based perturbation capable of destroying not only institutionalized values, but also the physical infrastructure and human capital of firms, this dissertation empirically demonstrates how political violence influences the strategies of MNEs. Furthermore, my interdisciplinary approach in integrating theoretical lenses from climate change and natural environment sustainability with existing management literatures to examine the effect of war on firms serves to enhance our understanding of individuals and collectives in extreme conditions.