Negligence: the statutory causation test in application

The provisions in three jurisdictions

Three jurisdictions use a similar test for causation in negligence. These are s 5D of the Civil Liability Act 2002 (NSW), s 11 of the Civil Liability Act 2003 (Qld), and s 51 of the Wrongs Act 1958 (Vic).

Each provision splits causation into two elements. Factual causation asks if the negligence was a necessary condition of the harm. Scope of liability asks if it is appropriate for liability to extend to that harm. Queensland’s section refers to “breach of duty” rather than “negligence”.

Each jurisdiction places the onus on the plaintiff to prove causation facts, on the balance of probabilities (NSW s 5E, Qld s 12, Vic s 52). Each also excludes certain claims, including some workers compensation and dust disease claims (NSW ss 5A(2) and 3B, Qld s 5, Vic s 45). These exclusions aren’t discussed further here, but students must check a claim falls within the right Part first.

The provisions differ in two ways relevant to this article.

First, how they treat cases where the “but-for” test isn’t satisfied. NSW s 5D(2) and Qld s 11(2) call this an “exceptional case”. Victoria’s s 51(2) instead asks whether an “appropriate case” exists where factual causation “should be taken to” apply, despite negligence not being a necessary condition.

Second, how they assess what the plaintiff would have done. All three use a subjective test, based on all relevant circumstances. However, NSW s 5D(3)(b) and Qld s 11(3)(b) add that the plaintiff’s own after-the-event statements about what they would have done are inadmissible, unless against their interest. Victoria’s s 51 contains no equivalent rule.

Factual causation is a but-for question, and the plaintiff must prove it

In Adeels Palace, a gunman shot two men at a restaurant’s New Year’s Eve function. They sued the operator for inadequate security. The High Court did not decide breach; it allowed the appeal because causation was not proved.

The Court said factual causation under s 5D(1)(a) is the “but-for” test: would the harm have happened without the negligent act or omission? Strong v Woolworths called this “a statutory statement” of that test, and Wallace v Kam said it is “nothing more or less” than it.

Three points follow. First, a negligent omission need not be the sole cause. If it is one of several conditions that together were sufficient to cause the harm, s 5D(1)(a) is satisfied (Strong).

Second, possibilities are not proof. In Adeels Palace, there was no evidence that security staff would have stopped the gunman. Pointing to what “might have” happened does not prove factual causation.

Third, showing the harm was the very kind the duty aimed to prevent does not itself prove factual causation. The Court said this may still be relevant to scope of liability, but it is a separate question.

For omissions, the plaintiff must prove the probable course of events had the omission not occurred. In Strong, a woman on crutches slipped on a chip in a store’s unswept sidewalk sales area. She had to show that, with a cleaning system, the chip would probably have been found and removed before she arrived. The evidence did not show when the chip fell, but the joint reasons held she could still prove causation by reasoning from probabilities, and allowed her appeal.

When but-for is not met

Quick Read: When but-for is not met

In Adeels Palace, the High Court said the but-for test (asking “would the harm have happened but for the defendant’s act?”) is normally necessary under s 5D(1). The only exception is the undefined group of “exceptional cases” under s 5D(2).

The Court noted the Act does not define “exceptional case”. Whether s 5D(2) applies depends on whether established legal principles allow departure from the but-for test.

The Court suggested s 5D(2) might cover cases like Fairchild v Glenhaven Funeral Services Ltd, where mesothelioma plaintiffs were exposed to asbestos across several jobs. It did not decide how s 5D(2) would actually apply to such a case.

Later, in Strong, the Court added that it had not yet considered whether a negligent material increase in risk can satisfy causation under common law principles. It also said that where several factors may have contributed to harm, and their relative roles cannot be separated, satisfying s 5D(1)(a) may depend on the scientific or medical evidence available.

These statements date from 2012; the article does not cover later cases. They interpret the New South Wales statutory wording only. The Victorian “appropriate case” test and the Queensland provision were not considered.

Scope of liability is the normative question

In Wallace v Kam, the Court explained how causation works under s 5D. Factual causation is “entirely factual” -- it asks what actually caused the harm. Scope of liability is “entirely normative” -- it asks whether the defendant should be legally responsible for that harm.

If a case fits an established type, precedent answers the scope question. This is because s 5D “guides but does not displace common law methodology”.

In a new or unusual case, s 5D(4) requires the court to explain, using legal policy, whether responsibility should be imposed and why. The Court warned that vague ideas like “directness”, “reality”, “effectiveness” or “proximity” are rarely enough on their own. “Common sense” is also of limited use unless the court explains the thinking behind it.

Illustration: a failure to warn of two risks

In Wallace v Kam, a surgeon failed to warn of two risks: temporary nerve damage (which happened) and a one-in-twenty chance of permanent paralysis (which did not). The Court applied s 5D to this common law negligence claim.

The Court set out three scenarios for factual causation. If the patient would have had the surgery anyway, even knowing all risks, causation is usually not established. If the patient would have refused the surgery altogether, causation is established. If the patient would have delayed the surgery but still had it later, causation is also established, as long as the injury remained a possible but unlikely outcome.

Because this turns on what the patient would have done, s 5D(3) applies. That provision excludes the patient’s own later statements about what they would have done, unless the statement is against their own interest.

On the facts, the patient would have refused surgery if warned of the paralysis risk. So factual causation was established for that risk.

Despite this, the Court held the surgeon should not be liable for the nerve damage. This is because the duty to warn exists to protect patients from risks they find unacceptable, and the patient would have proceeded with surgery if only warned of the nerve-damage risk.

The Court distinguished two situations. First, a single injury caused by several risk factors, where failing to warn of one factor may understate the overall risk. Second, several distinct risks of different injuries, where only one risk eventuates. The Court’s attribution rule—limiting liability to the warned-about risk—was designed for this second situation.

Applying this in a problem question

Start by finding the jurisdiction and reading its provision. The passage only covers New South Wales, Queensland and Victoria, and Victoria’s wording differs from the other two.

Next, state the harm and the specific negligent act or omission. Then ask the “but-for” question (would the harm have happened anyway?) on the balance of probabilities. The plaintiff bears the onus of proving this.

For an omission, say what the precaution would probably have achieved, with evidence for that. A mere possibility is not enough, following Adeels Palace. Probabilities can sometimes fill a gap in evidence, as the majority found in Strong (Heydon J dissented).

Only if but-for causation fails should you consider the statutory exception for an “exceptional” or “appropriate” case. Note that Adeels Palace discussed only the New South Wales wording, not a general test for when the exception applies.

If factual causation is established, scope of liability is a separate question. For an established class of case, use precedent; for a novel case, give an explicit policy explanation. In Wallace v Kam, the High Court said “directness” is rarely a sufficient test, and “common sense” has limited use unless its basis is explained.

For failure-to-warn cases, ask what this particular patient would have done. Apply the evidentiary rule about the plaintiff’s later statements, which exists in New South Wales and Queensland but not Victoria. Then consider scope of liability, remembering Wallace v Kam involved distinct risks causing different injuries.

Where the authority sits

This section explains which statute versions and cases are used. The New South Wales text is current from 30 July 2026. The Queensland text is in force as at October 2026. The Victorian text is Authorised Version No 130, with amendments as at 25 February 2026.

Adeels Palace, Strong and Wallace v Kam are High Court decisions on the New South Wales provision. Being High Court decisions, they bind lower courts, but only to the extent they actually decided a point. The High Court can still reconsider its own past decisions.

In Wallace v Kam the Court said s 5D is “substantially replicated” in other States and the ACT. But this passage does not set out those other jurisdictions’ wording. It only states the Queensland and Victorian provisions actually discussed.

So a student applying these cases outside New South Wales should first check the local wording themselves.

A Quick Read uses simpler language and may leave out detail, including whole topics, to stay short. The full article is the complete statement.

Read the full article →

Pop quiz

5 quick questions on this article, the authorities it cites and the articles it links to.

  • About 3 minutes, and no time limit.
  • You can only go forwards: each answer locks when you submit it.
  • After each question you see the right answer, why, and where to read more.
  • Free, and no account needed. Log in or create a free account to keep your scores.