Homeland security against possible terrorist attacks involves making decisions under true uncertainty. Not only are we ignorant of the form, place, and time of potential terrorist attacks, we are also largely ignorant of the likelihood of these attacks. In this paper, we conceptualize homeland security under true uncertainty as society's immunity to unacceptable losses. We illustrate and analyze the consequences of this notion of security with a simple model of allocating a fixed budget for homeland security to defending the pathways through which a terrorist may launch an attack and to mitigating the damage from an attack that evades this defense. In this problem, immunity is the range of uncertainty about the likelihood of an attack within which the actual expected loss will not exceed some critical value. We analyze the allocation of a fixed homeland security budget to defensive and mitigative efforts to maximize immunity to alternative levels of expected loss. We show that the production of homeland security involves a fundamental trade-off between immunity and acceptable loss; that is, for fixed resources that are optimally allocated to defense and mitigation, increasing immunity requires accepting higher expected losses, and reducing acceptable expected losses requires lower immunity. Greater investments in homeland security allow society to increase its immunity to a particular expected loss, reduce the expected losses to which we are immune while holding the degree of immunity constant, or some combination of increased immunity to a lower critical expected loss.