Fiscal multiplier literature uses war episodes as exogenous variation to achieve identification. However, the economy behaves differently in war episodes vis-à-vis peace periods; and hence the multiplier obtained using war episodes may be different from the multiplier prevailing in peace periods. To assess this assertion we use a calibrated New Keynesian model. We use these model to assess how different multipliers are in times of war and peace. We then turn to a Markov switching model in which government spending process is regime dependent, and we take into account this dependence when solving the model. We can obtain a difference of order four between war and peace multipliers using different definitions. In addition, we obtain a similar difference when we allow the presence of news in the model. These results shed light on the possible bias of the multipliers presented in fiscal literature.