But-for causation is the factual half of the causation element in a negligence claim: but for the defendant's conduct, would the plaintiff's injury have occurred? If the injury would have happened anyway, the defendant's conduct is not a cause-in-fact of it, and the causal chain never gets built far enough to reach proximate cause at all.

This is a factual, scientific, sequence-of-events inquiry -- it does not ask whether the harm was a foreseeable or fair consequence of the conduct. That policy-laden question belongs to proximate cause. Confusing the two is one of the most common analytical errors in negligence litigation, and courts are careful to keep them separate even though both use the word 'cause.'

The but-for test is counterfactual: remove the defendant's conduct from the sequence of events and ask whether the same harm still results. If the harm still happens, the conduct was not a cause-in-fact. If removing the conduct means the harm would not have happened when and how it did, cause-in-fact is established.

Plaintiffs typically must prove but-for causation by a preponderance of the evidence -- more likely than not, not to a scientific certainty. Expert testimony is frequently required in medical, toxic-tort, and product cases where the causal mechanism is not something a lay juror can assess unaided.

The but-for test breaks down when two or more independently sufficient causes combine to produce a single injury -- for example, two separate fires, each independently sufficient to destroy a building, that merge before reaching it. Applying but-for literally to each fire in isolation would let both defendants escape liability, since removing either one alone would not have prevented the loss.

Most jurisdictions respond with the substantial-factor test as a supplement (not a replacement) for but-for causation in these multiple-sufficient-cause scenarios: a defendant's conduct is a cause-in-fact if it was a substantial factor in bringing about the harm, even though it was not a necessary condition given the other sufficient cause present.

Where multiple defendants' negligent acts combine and it is unclear which one actually caused the injury -- the classic example being simultaneous hunters firing negligently where only one shot could have struck the plaintiff -- some courts shift the burden of proof to the defendants to disprove causation, rather than leaving an innocent plaintiff unable to identify the responsible party. Market-share liability in some toxic-tort and pharmaceutical contexts is a further, more specialized extension of this burden-shifting logic.

These doctrines remain the exception, not the rule; the ordinary case still requires the plaintiff to prove but-for causation against each named defendant.

In Juricratic, but-for causation is modeled as a distinct decision node from proximate cause, deliberately kept separate so a simulation cannot silently collapse the factual question (did this conduct actually produce the harm) into the legal-policy question (should liability extend this far). Counsel can dial the strength of the causal-chain evidence -- expert reliability, temporal proximity, alternative explanations -- independently of the proximate-cause scope dial, and see how each moves the exposure distribution on its own.

This separation matters most in multi-defendant and toxic-exposure fact patterns, where the causation fight is really about which test applies (but-for versus substantial-factor) before it is about the underlying facts at all.