In 1908 Mary Schloendorff agreed to be examined under ether at a New York hospital but, by her account, told the doctors not to operate. They removed a tumor anyway. Her lawsuit reached New York's highest court, where Judge Benjamin Cardozo wrote in 1914: "Every human being of adult years and sound mind has a right to determine what shall be done with his own body; and a surgeon who performs an operation without his patient's consent commits an assault, for which he is liable in damages." In other words, Cardozo called surgery without consent an assault. That sentence still anchors the idea that the patient, not the doctor, has the final say.

What consent requires

Agreeing is not enough. The phrase "informed consent" entered American law in a 1957 case, and ethicists now usually list several elements. The patient must have the capacity to decide. The clinician must disclose the relevant facts: the diagnosis, the proposed treatment, its risks and benefits, and the alternatives, including doing nothing. The patient must understand what was said. The choice must be voluntary, which means a patient who agrees under threat or heavy pressure has not given voluntary consent. Finally the patient must actually authorize the treatment. A signature on a form is evidence of consent, not consent itself.

How much must be disclosed?

Courts have used three standards. The older professional standard asks what a reasonable doctor would have said. In Canterbury v. Spence (1972), a federal appeals court in Washington, D.C., rejected that approach and adopted a patient-centered test: disclose what a reasonable patient would want to know to make the decision. A third, subjective standard asks what this particular patient would have wanted to know. The United Kingdom kept the professional standard for decades until the Montgomery case of 2015, when the UK Supreme Court required doctors to tell patients about any material risk a reasonable person in the patient's position would care about. The case involved a diabetic mother who was not told about a known risk of vaginal birth with a large baby.

Capacity

Adults are presumed to have capacity. Clinicians who doubt it usually check four abilities: to communicate a choice, to understand the information, to appreciate how it applies to one's own situation, and to reason about the options. Capacity is decision-specific: a person with early dementia may be able to choose a hospice but not to weigh a complex experimental treatment. An unwise choice is not proof of incapacity. If it were, the right to refuse would mean only the right to agree.

Deciding for others

When a patient lacks capacity, someone else must decide. Many places rank family members as default surrogates. The substituted judgment standard asks the surrogate to choose what the patient would have chosen, using the patient's own values, past statements or an advance directive. When nothing is known about the patient's wishes, as with infants, the surrogate follows the best interests standard and asks what would most benefit the patient. The two can diverge sharply: a devout patient might have wanted treatment a stranger would think pointless.

Paternalism and its limits

Paternalism means overriding someone's choice for their own good. Philosophers distinguish two kinds. Soft paternalism intervenes only when a choice may not be truly informed or voluntary, as when a confused patient tries to pull out a breathing tube. Hard paternalism overrides even an informed, voluntary choice, and modern medical ethics in most democracies rejects it for competent adults. A narrow exception called therapeutic privilege allows a doctor to withhold information only when disclosure itself would seriously harm the patient, not merely upset them. Courts treat it with suspicion because it is easy to abuse.

Critics of a strongly individualist model point out that in many cultures, and in many families everywhere, serious decisions are made together. A grandmother may want her children to hear a cancer diagnosis first and to decide with her. Defenders of autonomy answer that respecting her means asking her how she wants decisions made, not assuming either answer. Relational accounts of autonomy try to capture this: people choose as members of families and communities, and supporting their choices can mean supporting those relationships.

Worked example: refusing blood

A forty-year-old woman who is a Jehovah's Witness arrives at an emergency room bleeding after a car crash. She is alert, understands that she may die without a transfusion, and refuses blood on religious grounds. She carries a signed card saying the same. The surgeon thinks her choice is tragic.

The analysis is fairly clear. She has capacity for this decision: she understands, appreciates her situation, reasons from her values and communicates a choice. Her refusal is informed and appears voluntary, though staff may gently check, privately, that no one is pressuring her. Overriding her would be hard paternalism and, under the Schloendorff principle, an assault. The team gives her every blood-free treatment it can, such as fluids, drugs that reduce bleeding and careful surgery. If she loses consciousness, her card guides them as an advance directive. The case changes when the patient is a child. Courts in many countries routinely authorize transfusions for minors over parents' objections, reasoning that a child has not yet chosen a faith and that parents' rights do not include letting a child die.