We develop a model of investments in interdependent security systems in the presence of a strategizing opponent, who attempts to infiltrate at least one of the systems with a harmful device. In the first stage, a finite number of targets decide about their respective investments in direct and indirect security systems by choosing appropriate detection probabilities minimizing expected losses. In the second stage, infiltration of a target’s security system with the harmful device is directed by the agent from the outside and may occur either directly by a target’s failing to detect the device at its entry level or indirectly by first passing undetected through the other target’s entry screening and subsequently clearing the cross-screening device.