We conducted an economic lab experiment to investigate different framing-effects in a game-theoretic defender-attacker model. Adversarial risks, such as crime and terrorism, are some of the biggest challenges for decision-makers in business and politics these days. Nevertheless, when analyzing them, decision-makers very often have the problem that information for testing and validating their theoretical models is not available, because information is either confidential or not existent. To be able to test, refine, and compare theoretical models, researchers must thus find other ways to generate reliable information. In the field of economics, one such way is to carry out lab experiments. In lab experiments, scientists describe reality in an abstract and artificial way, to test critical elements of a theory. However, so far, scientists conducted only a small number of lab experiments in the context of adversarial risks compared to the number of existing theoretical models. These experiments are usually concerned with interdependent security games, Colonel Blotto games, punishment and vendettas, global security games, or the trade-off between civil liberties and national security. All these experiments on adversarial risks have in common that they do not address the effects of framing on the strategy selection of the participants. Framing is a cognitive bias, which causes people to make different decisions depending on how a decision problem is presented to them, e.g., whether they are expecting gains or losses. To this end, we developed a generic game-theoretic defender-attacker model with continuous strategy-sets, for which we tested the Nash equilibrium strategies under different framing-conditions. Since adversarial risks represent severe conflicts of interests, we assume that framing has particular strong effects on the participants' decisions. In our experiment, we investigated two framing-effects: framing through payoffs (the participants received either gains or losses), and framing through wording (we described the games either in a neutral or in an aggressive way). The main results of our experiment are that the participants, in particular participants who had been assigned the role of the defender, behaved similar as predicted by the model, and that framing through payoffs and/or wording can have significant effects on the participant’s strategies.