Fundamentals

Harm Principle: What It Is and Its Limits

By Daniel Sardá · Published on September 23, 2026

6 min read1,132 words

The harm principle begins with a presumption of liberty: restricting a person requires showing a significant injury to others, not merely disapproval or a wish to protect them from themselves.

The harm principle holds that coercion against a person can generally be justified only to prevent harm to others. It is not enough to think that a person's conduct is imprudent, immoral, or contrary to their own welfare.

Associated with John Stuart Mill and his essay On Liberty, this idea is a way of thinking about the limits of power, not a rule of tort law or an automatic answer to every conflict. Its central question is simple but demanding: what must someone show before restricting another person's freedom?

Key idea: Liberty is the starting point. The burden of giving reasons falls on the person proposing a prohibition, a penalty, or another form of coercion.

Mill's formulation: preventing harm, not imposing the good

Mill published On Liberty in 1859. In its first chapter, he sets out the core of the principle: power may legitimately be exercised against someone's will only to prevent harm to others; that person's own physical or moral good is not, by itself, a sufficient reason. The formulation can be consulted in the original text of *On Liberty*.

The distinction matters because a person can make a bad decision without thereby injuring another person's rights or important interests. In that case, one may advise, warn, or try to persuade them. What needs additional justification is compelling them through penalties, prohibitions, or organized social pressure.

Mill was not concerned only with the state. He also worried about majorities' capacity to impose their customs and opinions. The principle therefore protects a sphere of civil liberty against law and against social coercion.

Harm to others and self-regarding conduct

Conduct is self-regarding when its consequences fall principally on the person acting. The clearest case might be a private activity among informed adults who freely consent and do not shift significant costs or risks onto other people. Even if outside observers consider it unwise, disagreement does not establish that harm exists.

The situation changes when affected third parties enter the picture. If a decision breaches a specific duty to a dependent person, exposes others to a significant danger, or rests on fraud or coercion, it can no longer simply be described as a private choice. One must identify who is affected, what interest is injured, and through what causal connection.

Consent also has conditions. It does not perform its role when obtained through deception or threat, or when the person lacks the capacity to understand the decision. Mill also excluded minors from the full scope of his principle. These qualifications prevent autonomy from becoming a magic word that ends every discussion.

Harm is not the same as offense

Feeling annoyed, scandalized, or offended does not necessarily amount to being harmed. A plural society requires a degree of tolerance toward other people's conduct, precisely because people disagree about religion, morality, ways of life, and conceptions of the good.

The distinction is not always comfortable. Public expression may cause intense displeasure and, in other settings, similar conduct may intimidate, harass, or obstruct the exercise of a right. That is why cases cannot be classified solely by the emotion they provoke. It is better to ask whether an important interest has been injured, whether there is an identifiable victim, and whether the effect goes beyond mere disapproval.

Scholarly interpretations of Mill commonly connect harm with important interests and rights, while acknowledging tensions around certain public offenses. The Stanford Encyclopedia of Philosophy provides an overview of those debates.

Useful distinction: Offense describes a reaction of displeasure; harm requires showing a significant injury or threat. Confusing the two makes it possible to turn any majority preference into a reason to coerce.

Risk: there is no need to wait for injury

The principle does not require passively waiting until harm occurs. There may be reasons to prevent conduct that creates a defined risk for others. But invoking the word “risk” is not enough: its probability, severity, proximity, and empirical support matter.

Consider three hypothetical cases. An unhealthy household activity that affects only the adult who chooses it primarily raises a problem of self-harm. The same activity carried out in a way that exposes neighbors to a dangerous substance brings third parties into the picture. And an activity that produces only an occasional nuisance may call for accommodation or narrowly tailored rules, but not necessarily a general prohibition.

These examples do not dictate universal legal outcomes. They help organize inquiry: specify the danger, compare scenarios, and avoid treating a remote possibility as if it were a grave and immediate threat.

Borderline cases: indirect harms and paternalism

Many decisions have some social repercussion. If every indirect effect counted as sufficient harm, almost no conduct would remain outside collective control. To avoid that expansion, a claim of harm should specify at least the victim or affected group, the interest at stake, the causal connection, and the expected magnitude.

It is also useful to separate protecting others from paternalism: interference with someone's will in order to protect or benefit them. “Soft” paternalism seeks to establish whether a decision is truly voluntary and informed; “hard” paternalism permits intervention even when it is. This distinction, developed in the Stanford Encyclopedia of Philosophy entry on paternalism, shows that not every protective measure rests on the same argument.

There are also collective, cumulative, or difficult-to-attribute harms. The principle does not make them unreal, but it requires careful argument. The more diffuse the causal chain, the greater the need for evidence and institutional limits that prevent a vague injury from becoming unlimited authorization.

Showing harm is necessary, but not always sufficient

Even where there is a significant injury to others, one must still ask whether the proposed intervention is proportionate, effective, and less harmful than its alternatives. Information, compensation, a more precise rule, or a limited preventive measure may protect those affected better than a broad prohibition.

This is where the harm principle connects with limits on coercion. Institutions make mistakes, may apply rules unevenly, and generate costs of their own. Evidence of harm therefore opens the discussion about a restriction; it does not give a blank check for any response.

Practical test: Before restricting conduct, identify the harm to others, show a plausible connection, and explain why the chosen measure is proportionate and preferable to less restrictive options.

The harm principle remains powerful because it does not promise to eliminate every disagreement. It demands something more modest and more difficult: distinguishing moral judgment from legitimate coercion. In a free society, we may criticize other people's choices, seek to persuade them, and demand responsibility for actual harms. What we may not do without sufficient reasons is turn every disapproval into an order.