Proximate Cause in Torts: Foreseeability, Palsgraf, and Scope of Liability

Proximate cause asks whether a defendant should be held liable for a harm that its negligence in fact caused. It is a policy limit on liability, not a factual inquiry, and modern treatments increasingly call it scope of liability for that reason.

It operates only after cause in fact is established. Cause in fact asks whether the harm would have occurred absent the defendant's conduct — the but-for test, supplemented by the substantial-factor test where multiple sufficient causes are present. If the answer is no, the analysis ends and proximate cause is never reached.

The reason a second filter exists is that but-for causation extends indefinitely. Every negligent act sits at the head of an unbounded chain of consequences, and without a limiting principle a driver's momentary inattention could ground liability for harms decades removed. Proximate cause is where courts draw that line.

The dominant approach asks whether the harm was a reasonably foreseeable result of the defendant's negligence. The competing direct-cause approach asks whether the harm followed from the negligent act without an intervening force sufficient to break the chain, regardless of foreseeability.

The foreseeability test traces to Overseas Tankship (U.K.) Ltd. v. Morts Dock & Engineering Co. — The Wagon Mound (1961) — where oil negligently discharged into a harbor ignited and destroyed a wharf. The court held the fire damage unforeseeable and denied recovery, displacing the direct-cause approach of In re Polemis. Most American jurisdictions apply some version of foreseeability today.

The Third Restatement reframes the test as whether the harm resulted from the risks that made the actor's conduct tortious. This formulation is often clearer on exams because it directs attention to the specific reason the conduct was negligent. If a pharmacy negligently dispenses the wrong medication and the patient is injured by the drug, the harm is within the risk. If the patient trips on the curb leaving the pharmacy, it is not — even though but-for causation holds.

In most cases the tests converge. They diverge in the unusual manner-of-harm and extent-of-harm cases below, which is why those cases dominate casebooks.

Palsgraf v. Long Island Railroad Co. (1928) is taught as the central proximate cause case even though Cardozo's majority opinion resolves it as a question of duty. Railroad employees helping a passenger board dislodged a package of fireworks, which exploded, and the concussion allegedly toppled scales at the far end of the platform onto Helen Palsgraf.

Cardozo held there was no negligence toward Palsgraf at all: negligence is relational, and a duty runs only to those within the zone of foreseeable danger. Because no reasonable person would anticipate harm to a person standing that far away from the act of steadying a passenger, she had no cause of action to begin with. "The risk reasonably to be perceived defines the duty to be obeyed."

Andrews, dissenting, argued that a duty of care runs to the world at large, and that whether to impose liability on these facts is a proximate cause question — to be answered with reference to a set of practical considerations including the directness of the connection, the remoteness in time and space, and whether the result was too attenuated.

The framing matters for exam structure. Under Cardozo, an unforeseeable plaintiff loses at the duty element; under Andrews, the same facts are analyzed at proximate cause. A strong answer notes which framing the jurisdiction follows and then applies it, rather than treating the two as interchangeable.

The eggshell skull rule is the most frequently tested row. A defendant who negligently causes a minor impact is liable for the full extent of injury even where a rare pre-existing condition turns a bruise into a catastrophic outcome. The type of harm — physical injury from impact — was foreseeable; the magnitude need not be.

The rescuer rule from Wagner v. International Railway Co. is the second. Because it is foreseeable that someone will attempt rescue when a defendant's negligence places another in peril, the rescuer is owed a duty and injuries sustained in a reasonable rescue attempt are within the scope of liability.

An intervening cause is any force that contributes to the harm after the defendant's negligent act. A superseding cause is an intervening cause that is unforeseeable enough to cut off the defendant's liability. Every superseding cause is intervening; most intervening causes are not superseding, and the distinction is the entire analysis.

The controlling question is whether the intervening force was itself within the range of foreseeable consequences. Ordinary negligence by a third party is generally foreseeable and does not supersede. Medical negligence in treating an injury the defendant caused is the standard example: the original tortfeasor remains liable for aggravation caused by ordinary malpractice, because seeking treatment is a foreseeable consequence of being injured.

Intentional torts and criminal acts by third parties are more often superseding, but not automatically. When the defendant's negligence consisted precisely in failing to guard against such conduct — a landlord who fails to repair a broken lock, a hotel with no security in a high-crime area — the criminal act is the very risk that made the conduct negligent, and liability follows.

Acts of God supersede when extraordinary and unforeseeable. A routine storm in a region where storms are common generally does not.

The framing worth remembering: an intervening cause supersedes when it is so unforeseeable that holding the defendant liable would extend responsibility beyond the risks that made the conduct wrongful in the first place.

Merging proximate cause with cause in fact. They are separate elements answering different questions, and an answer that establishes but-for causation and stops has skipped an element. Address them in order and label them.

Requiring the exact sequence of events to be foreseeable. The manner of occurrence need not be anticipated, only the general type of harm. An answer that denies liability because the specific chain was bizarre is applying the wrong standard.

Treating any third-party act as superseding. Foreseeable third-party negligence does not break the chain, and criminal conduct does not break it where preventing that conduct was the point of the duty.

Using the eggshell skull rule to establish liability rather than to measure damages. It does not make an unforeseeable type of harm compensable; it removes the extent of a foreseeable type of harm from the foreseeability inquiry.

Ignoring the Cardozo/Andrews split. On facts involving a remote plaintiff, identifying whether the issue is framed as duty or as proximate cause is often what separates a strong answer from an adequate one.