Fundamentals

The Non-Aggression Principle: What It Is, What It Forbids, and Its Limits

By Daniel Sardá · Published on September 24, 2026

6 min read1,209 words

An explanation of the non-aggression principle, how it differs from pacifism, and the questions raised by property, fraud, and threats.

The non-aggression principle (NAP) holds that no one should initiate the use of force—or threaten to do so—against another person or that person's legitimate property. It is a moral rule associated with several libertarian traditions, though those traditions do not all interpret it in the same way or give it the same scope.

Its formulation seems simple: do not attack, steal, or coerce. The difficulty begins when it is applied. What counts as a threat? How much force may be used in self-defense? When does something belong legitimately to someone? Is fraud a form of aggression? The answers depend on prior ideas about property, consent, causation, and responsibility.

The NAP is therefore a starting point for examining voluntary relations and the limits of coercion. By itself, it is neither a complete legal code nor an automatic solution to every conflict.

Key idea: The core of the NAP is not “never use force,” but “do not be the one who initiates it.”

What does the non-aggression principle forbid?

In its best-known version, developed by libertarian writers such as Murray N. Rothbard, the principle forbids the physical invasion of a person or that person's property. It covers clear cases such as assault, kidnapping, theft, or the deliberate destruction of someone else's property.

The decisive idea is the initiation of force. If someone hands over a phone because another person threatens them with a weapon, the transfer is not voluntary: it is coerced. A heated argument or an offensive opinion, by contrast, is not in itself aggression in this sense, even if it may be objectionable for other reasons.

This distinction prevents every harmful, rude, or unpleasant act from being treated as an invasion of rights. To describe conduct as aggression, one must identify which right was violated, by what conduct, and against whom.

Non-aggression does not mean pacifism

The NAP allows, in principle, the use of force to stop an unjust aggression. It is therefore not the same as pacifism, which may reject violence even in self-defense.

The permission to defend oneself is not unlimited, either. The philosophy of self-defense commonly distinguishes at least two conditions: necessity and proportionality. The first asks whether force was necessary to stop the attack; the second asks whether the response was proportionate to the seriousness of the threat.

Pushing away someone who is trying to strike another person in order to stop the attack may be defensive. Pursuing that person hours later to inflict much greater harm raises a different question: it is no longer about stopping an immediate danger, but about retaliation or punishment. Restitution and sanctions also require their own criteria.

The same is true of threats. A vague warning or hostile expression is not the same as a credible and imminent threat. Rothbard defended a demanding threshold—a direct, overt, and immediate threat—but that is a doctrinal formulation, not a universal legal test.

Key idea: Self-defense seeks to stop aggression; retaliating afterward requires a different justification.

Property is part of the definition

The reference to “legitimate property” introduces the NAP's most important debate. To know whether someone has invaded property, one must first establish who had a right to it. The rule against aggression does not by itself supply a complete theory of acquisition, contract, inheritance, abandonment, or restitution.

The philosopher Matt Zwolinski identifies this difficulty in his critical assessment of the principle: if aggression is defined as a violation of property rights, it remains necessary to justify which property rights exist. Otherwise, the reasoning risks circularity: an action would be aggressive because it violates property, and that property would be valid because the NAP protects it.

This does not make the principle useless. It does require a distinction between its basic rule and the assumptions it needs. Two people may agree that theft is unjust while still disagreeing about the original entitlement to a resource or the conditions of a legitimate transfer.

That discussion connects the NAP with property rights, but the two should not be confused. The former limits the initiation of force; the latter helps determine which goods each person may control and from which others may be excluded.

Consent, contract, and fraud

Consent separates a voluntary exchange from a coercive taking. If two people understand the essential terms of an agreement and accept them without force or threat, the NAP provides no reason to prevent the exchange merely because a third party considers it unwise.

Fraud is a less straightforward case. On the Rothbardian view, obtaining an asset by deliberately promising something one does not intend to deliver can be treated as an implicit form of theft: the victim consented to a transfer different from the one that actually took place. But not every mistake, exaggeration, or breach of contract by itself establishes aggression. Intent, the agreed terms, and the causal relation between the deception and the loss all matter.

For that reason, it is wise to avoid absolute claims such as “every breach of contract violates the NAP.” A dispute about the quality of a service, an ambiguous promise, and a planned fraud are not the same problem.

Borderline cases reveal its limits

Clear cases explain the principle's intuition; difficult ones reveal what is missing for its application. The main problems include indirect harms, collective risks, emergencies, and conduct whose causal connection is uncertain.

Suppose an activity creates a minor nuisance that crosses a property boundary. It is not enough to say that some effect exists: its magnitude, cause, the rights at stake, and a reasonable remedy would all need to be specified. Nor does every emotional offense amount to a physical invasion, even if it may justify social criticism or responsibility under other rules.

Strict versions of the NAP try to resolve nearly all such cases through robust property rights. More moderate versions treat it as a strong presumption against coercion that can come into tension with other duties in exceptional circumstances. The Stanford Encyclopedia of Philosophy emphasizes that libertarianism includes diverse positions; there is no single libertarian answer to every conflict.

Key idea: Calling conduct “aggression” does not settle the debate if property, consent, causation, and proportionality have not yet been clarified.

Its place in classical liberalism

The NAP shares with classical liberalism a concern for individual liberty, property, and limits on power. They are not equivalent concepts, however. Classical liberalism includes historical and institutional arguments about limited government, the rule of law, tolerance, and open markets; the NAP is a more specific moral rule, associated mainly with contemporary libertarianism.

It is also important to distinguish the definition from its political consequences. Some doctrines conclude that many state actions constitute aggression; others accept limited coercive institutions to protect rights or resolve disputes. That controversy cannot be settled by repeating the principle: it requires an account of authority, property rules, and mechanisms of protection.

The NAP's value lies in posing a demanding but intelligible question: who initiated force, and on what justification? The question requires coercion to be explained rather than assumed. Its limit is equally important: without a theory of rights, responsibility, and proportionate response, the rule guides analysis but does not complete it.