The Security Dilemma

In brief

The security dilemma is the founding puzzle of international relations. Its logic: in a world with no government above governments, each state must provide for its own safety, and the measures it takes to do so, more weapons, new bases, tighter alliances, are indistinguishable from the measures a would-be aggressor would take. A neighbor watching the buildup cannot read intentions, only capabilities, so it arms in response. The first state, seeing this, feels its fears confirmed and arms further. Each acts defensively, each is rational, and the outcome is a spiral of mutual insecurity that no one wanted. The concept was named by John H. Herz in 1950 and given its modern analytical shape by Robert Jervis in 1978. It explains how genuine conflict, including war, can arise between states that have no aggressive designs on each other at all.

The full treatment

The problem it answers

Realist theory holds that the deepest fact about international politics is anarchy, meaning the absence of a common power, not chaos. Inside a state, a citizen who feels threatened can call the police and appeal to courts. Between states there is no equivalent: no world government, no enforceable law backed by a monopoly on force, no impartial arbiter to punish the aggressor and protect the victim. Each state is the final guarantor of its own survival. The security dilemma is what this condition does to states that only want to be left alone. It answers a question a simple story of good states and bad states cannot: how can there be arms races, crises, and wars even when every party sincerely prefers peace and seeks only to defend what it has?

How the spiral works

The engine is uncertainty about intentions. A state can observe what another state builds, but it cannot see into the minds of that state's leaders, and it cannot be sure their intentions will not change, or that today's cautious ruler will not be replaced by tomorrow's ambitious one. Since a mistake about a rival's aims can be fatal, prudent statecraft plans against capabilities rather than professed intentions. So State A, feeling insecure, acquires arms purely to defend itself. State B cannot distinguish A's defensive buildup from preparation for attack, and, unwilling to gamble its survival on A's good faith, arms in reply. A reads B's response as proof its original fear was justified, and redoubles its effort. The two climb a ladder of mutual arming that leaves both spending more and feeling less safe than before. The tragedy, in Herbert Butterfield's phrase, is that of "Hobbesian fear": the whole dynamic can run on fear alone, with no villain anywhere in the system.

Herz, Butterfield, and Jervis

The political scientist John H. Herz (1908 to 2005), a German emigre writing in the United States, introduced the term in a 1950 article in World Politics and developed it in Political Realism and Political Idealism (1951). He located the dilemma not in human wickedness but in the structural fact of living alongside others without a protector: the drive for security itself generates insecurity. The historian Herbert Butterfield (1900 to 1979) reached a strikingly similar idea in History and Human Relations (1951), describing how two powers can be locked in conflict by mutual fear despite both being, in their own eyes, entirely on the defensive.

The sharpest analytical development came from Robert Jervis (1940 to 2021). In Perception and Misperception in International Politics (1976) and above all in "Cooperation Under the Security Dilemma" (World Politics, 1978), Jervis argued that the dilemma's severity is not fixed but varies with two conditions. The first is the offense-defense balance: whether prevailing military technology and geography make attacking easier than defending, or the reverse. When offense has the advantage, small shifts in strength are dangerous, first strikes are tempting, and the dilemma bites hard. When defense has the advantage, states can be secure without threatening others, and the spiral slackens. The second condition is whether offensive and defensive weapons can be told apart. When they can (fortifications and coastal defenses, say, versus long-range bombers), a state can signal benign intent by choosing the defensive kind. When they cannot, every acquisition is ambiguous and fear compounds. Jervis turned a general insight into a set of variables that predict when cooperation is possible and when it is nearly impossible.

Distinctions that matter

Two distinctions keep the concept honest. First, the security dilemma is not ordinary conflict of interest. States sometimes genuinely want incompatible things, which produces conflict for straightforward reasons. The dilemma names something more unsettling: conflict between states whose interests do not actually clash, generated purely by the structure they inhabit. Jervis paired it with a warning against the "deterrence model," the assumption that firmness always deters. Under the spiral model, treating a fearful adversary as an aggressor and answering it with toughness confirms its fears and drives the escalation you meant to prevent. Reading which situation you are in, a real aggressor to be deterred or a frightened state to be reassured, is the central and often unanswerable problem of statecraft.

Second, the dilemma is distinct from the Hobbesian state of nature to which it is closely related. Thomas Hobbes (1588 to 1679) described individuals in a stateless condition arming against one another out of fear; the security dilemma applies that logic to states, where it is both more durable, since states do not die of natural causes and cannot all submit to a single sovereign, and partly softened, since states are less vulnerable than individuals and can survive misjudgments that would kill a person.

Lineage

The idea descends most directly from the Hobbesian analysis of life without a common power, transposed from persons to states. It draws on a long realist tradition running back to Thucydides (c. 460 to c. 400 BCE), whose account of the Peloponnesian War traced the conflict to Spartan fear of growing Athenian power, a fear that generated war between polities that might otherwise have coexisted. In the twentieth century the concept became a load-bearing element of structural realism, the school built by Kenneth Waltz (1924 to 2013) in Theory of International Politics (1979), which grounds state behavior in the anarchic structure of the system rather than in the nature of individual leaders. Herz supplied the mechanism, Jervis the variables, and later scholars including Charles Glaser, Stephen Van Evera, and the pair Ken Booth and Nicholas Wheeler (whose 2008 book stressed how much rests on interpretation and the possibility of trust) refined and contested it.

The strongest case for it

The security dilemma explains a class of events that theories of good and evil states cannot. Arms races such as the Anglo-German naval competition before 1914, crises that escalate faster than either side wants, and wars both parties later regret all become intelligible once you see that fear alone, working through anarchy and uncertainty, is sufficient to produce them. The concept is parsimonious: it derives large consequences from a single structural fact, without assuming aggression, irrationality, or malice. It is also action-guiding. Jervis's offense-defense framework yields concrete advice: build recognizably defensive forces, favor arms-control regimes that let states verify one another's capabilities, and design postures that do not reward striking first. Much of the logic behind arms control and confidence-building measures during the Cold War rests on exactly this reasoning. Its refusal to moralize is itself a strength: by locating the danger in structure rather than character, it warns that even a world of well-intentioned leaders is not safe, a more demanding lesson than the hope that peace follows from good intentions.

The strongest case against it

The concept has real critics who attack it at the root. The most influential challenge comes from constructivism. Alexander Wendt (born 1958), in "Anarchy Is What States Make of It" (International Organization, 1992) and later work, argued that anarchy has no fixed logic. Whether it breeds a security dilemma depends on the shared ideas and identities states bring to their relations. The United States does not fear British nuclear weapons but does fear far smaller North Korean ones, which shows that capabilities alone do not generate the spiral; the meaning attached to them does. On this view the dilemma is not a permanent structural fact but a self-fulfilling belief that fades when states come to see one another as friends rather than potential enemies, as the members of the European Union largely have.

A second line of attack comes from within realism. Defensive realists such as Charles Glaser (born 1954) argue that the standard account overstates how often the dilemma actually bites: when defense has the advantage and weapons are distinguishable, a security-seeking state can often signal its intentions credibly and avoid the spiral. A related worry is that the concept can be misused as an all-purpose excuse, since a genuinely expansionist state can always claim its buildup is merely defensive, and observers who reach too quickly for the security-dilemma explanation may fail to deter a real aggressor, precisely the error Jervis's deterrence model warns against. Liberal theorists add that the account ignores the forces that dampen the spiral in practice: international institutions that make intentions legible, economic interdependence that raises the cost of conflict, and domestic regime type, since democracies rarely fight one another. Where those forces are strong, anarchy need not produce the tragic outcome the theory predicts.

Where it stands now

The security dilemma remains one of the most widely used concepts in international politics, taught in nearly every introductory course. It is central to debates about the rise of China and the risk of great-power war, where analysts ask whether Chinese and American military moves are locking the two states into a spiral neither wants, a version of the pattern sometimes called the Thucydides trap. It informs analysis of nuclear postures, missile defense (which can look defensive to its builder and destabilizing to its rival), and the eastward expansion of alliances. The constructivist challenge has been absorbed rather than defeated: most scholars now accept that the dilemma is conditional, its intensity shaped by technology, geography, institutions, and above all by how states interpret one another, rather than an iron law of anarchy. That refinement made the concept more careful without retiring the hard insight at its center, that the pursuit of security can itself be a source of war.

Test yourself

Think of a rivalry you consider one-sided, where one side is clearly the aggressor and the other clearly on the defensive. Now try to describe the same events from the other side's point of view, assuming it too believes it is only defending itself. If that account is even coherent, you have found a security dilemma, and you have also felt why it is so hard, from inside a crisis, to know whether you are facing an aggressor to deter or a frightened rival to reassure.