Public concern about soldiers' mental health has increased over the last decade. Yet the large literature on the mental health problems of returning soldiers relies primarily on self-reported measures that may suffer from non-response bias, usually refers to older conflicts, and focuses mainly on specific diagnoses such as PTSD. Another challenge is that the differences between soldiers and non-soldiers are not necessarily causal, instead possibly reflecting an underlying propensity towards active military service. Using the objective measures of hospitalizations and the purchase of mental health medication, this paper is the first to investigate the effect of recent military deployments on a broader measure of mental health, for a full population of Danish soldiers and a comparison group of eligible men. We exploit a panel of Danish health administrative records and use propensity score matching process, using unique observables such as ability tests. Matching is helpful in both reducing the selection bias from observables and determining pre- and post-deployment period for the comparison group. Then, we estimate the effect of deployment in a difference-in-differences setting, controlling for time trends and other omitted variables affecting both groups. Overall, we find a significant and long-lasting adverse effect of military deployment on soldiers' mental health, regardless of the comparison groups and underlying assumptions.