Few areas of private law have been remade as often as the duty of care in negligence. In the nine decades since Donoghue v Stevenson (1932), Australian courts have adopted, refined and ultimately discarded a succession of general formulas purporting to explain when one person must take reasonable care to avoid injuring another, before settling on the frankly pluralist method that prevails today. This essay traces that evolution and defends a qualified thesis: the modern Australian approach, an incremental method organised around the salient features catalogued in Caltex Refineries (Qld) Pty Ltd v Stavar (2009) and disciplined by the coherence principle articulated in Sullivan v Moody (2001), is more intellectually honest than the unifying tests it replaced, although it purchases that honesty at some cost to predictability. The most serious incoherence in contemporary negligence law lies not in the common law method itself but in its fragmented interaction with the state and territory civil liability legislation.
The argument proceeds in five stages: the reception of the neighbour principle into Australian law; the rise and collapse of proximity in Jaensch v Coffey (1984) and its successors; the High Court’s turn to coherence in Sullivan v Moody; the consolidation of the salient features approach; and the statutory reforms that followed the Ipp Review. Against that history, the final section evaluates how Australian courts now confront novel duty categories and whether the resulting law merits the description coherent.
The Neighbour Principle and its Australian Reception
Modern negligence law begins with Lord Atkin’s celebrated generalisation in Donoghue v Stevenson (1932) that reasonable care must be taken to avoid acts or omissions likely to injure one’s neighbour, meaning persons so closely and directly affected by the conduct that they ought reasonably to be within the defendant’s contemplation. The neighbour principle was never a complete test of liability. It was a generative idea, a moral rationalisation from which particular duties could be derived (Plunkett 2018, p. 34). Its immediate doctrinal work was to release the manufacturer’s liability from the grip of privity of contract; its longer significance lay in supplying negligence with a unifying aspiration.
The Australian reception was swift and distinctly local. In Grant v Australian Knitting Mills Ltd (1936), an Adelaide physician contracted severe dermatitis from sulphite residues left in woollen underwear manufactured in South Australia. On appeal from the High Court of Australia, the Privy Council applied Donoghue and confirmed that a manufacturer owes a duty to the ultimate consumer, anchoring the neighbour principle in Australian law (Luntz et al. 2021, p. 141). For physical injury caused by positive acts, reasonable foreseeability quickly became, in practice, the only question, and duty was rarely contested within established categories (Stickley 2016, p. 204).
The difficulty was that foreseeability is an undemanding criterion. Almost any harm is foreseeable at some level of generality, so foreseeability alone could not control liability for psychiatric injury, pure economic loss, negligent omissions or the exercise of statutory powers by public authorities. The subsequent history of the duty of care in Australia is best read as a prolonged search for the additional element that separates foreseeable harms attracting obligation from those that do not (Stewart & Stuhmcke 2020, p. 168).
Proximity and the Search for a Unifying Principle
The High Court’s most ambitious answer was proximity. In Jaensch v Coffey (1984), the wife of a South Australian police motorcyclist developed a psychiatric illness after attending hospital and observing her husband’s injuries in the aftermath of a collision caused by the defendant’s negligent driving. The Court allowed her claim, and Deane J used the occasion to elevate proximity, comprising physical, circumstantial and causal closeness between the parties, into the conceptual determinant of the duty of care. On this view, foreseeability marked the outer boundary of negligence while proximity explained why particular relationships fell inside it, and for roughly a decade the concept operated as the Court’s organising principle (Sappideen, Vines & Watson 2016, p. 178).
Proximity nevertheless failed to yield a workable test. The persistent criticism was that it described conclusions rather than produced them: a court that had already decided a duty should exist labelled the relationship proximate, so the concept concealed rather than exposed the true grounds of decision. By Hill v Van Erp (1997), where a solicitor was held to owe a duty to an intended beneficiary deprived of a legacy by careless will drafting, majority support had visibly ebbed. In Perre v Apand Pty Ltd (1999) the concept collapsed. Apand had introduced bacterial wilt onto a farm neighbouring the Perre families’ South Australian potato properties, and because Western Australian regulations excluded potatoes grown within twenty kilometres of an outbreak, the Perres lost access to the lucrative Perth market even though their own crops were untouched. The Court unanimously recognised a duty in respect of this pure economic loss, but did so through seven separate judgments deploying incrementalism, multifactorial analysis and, in Kirby J’s case, the English three stage test from Caparo Industries plc v Dickman (1990). Proximity was abandoned as a legal test, surviving only as a loose description of the connections that may matter.
Sullivan v Moody and the Turn to Coherence
The resulting methodological vacuum was confronted in Sullivan v Moody (2001). Medical practitioners and social workers who had examined children in South Australia for suspected sexual abuse were sued by the fathers under suspicion, who alleged that negligent investigation had caused them psychiatric injury and the destruction of family and reputation. A unanimous High Court held that no duty of care was owed. The examiners’ statutory function under the child protection legislation required them to treat the interests of the children as paramount; a common law duty to the suspects would pull in the opposite direction, and a duty organised around reputational harm would subvert the calibrated balance struck by defamation law.
Two features of the judgment shaped everything that followed. First, the Court expressly rejected both proximity and the Caparo formula as governing tests, observing that different classes of case present different problems which no single verbal formula can resolve. Secondly, it installed coherence as a controlling consideration: a duty will not be recognised where it would contradict other legal obligations of the defendant or distort adjoining bodies of law. Coherence has obvious normative appeal, because a legal system should not command incompatible behaviour, yet its operation has proved conspicuously asymmetric. It is invoked almost exclusively to deny duties and rarely to support them, giving the modern law a restrictive tilt that is itself a distributional choice requiring justification (Dietrich 2020, p. 14).
The Salient Features Approach
With unifying theory abandoned, Australian law still needed a transparent method for novel cases, and it was supplied by the New South Wales Court of Appeal in Caltex Refineries (Qld) Pty Ltd v Stavar (2009). Mrs Stavar, who for years had washed her husband’s asbestos contaminated work clothing, developed mesothelioma and sued the refinery operator. Whether such an operator owes a duty to an employee’s spouse was novel, and Allsop P answered it by cataloguing some seventeen salient features bearing on the existence of a duty, including foreseeability, the defendant’s control over the risk, the plaintiff’s vulnerability, reliance, assumption of responsibility, the spectre of indeterminate liability, and consistency with statutory schemes and contractual arrangements. The features are not a checklist to be tallied. They structure an evaluative judgment in which foreseeability is necessary but never sufficient, and no single feature is decisive (Robertson 2019, p. 520).
Although the High Court has never adopted the Stavar catalogue in terms, its own reasoning tracks the same considerations. In Brookfield Multiplex Ltd v Owners Corporation Strata Plan 61288 (2014), a builder owed no duty to a subsequent owners corporation for the cost of repairing latent defects in a Sydney serviced apartment development, principally because the sophisticated contractual framework demonstrated that the purchasers were not vulnerable: they could have protected themselves by bargaining. Earlier, in Graham Barclay Oysters Pty Ltd v Ryan (2002), the limited control exercised by the State of New South Wales over the Wallis Lake catchment defeated a claim that the government owed a duty to consumers who contracted hepatitis A from contaminated oysters. Vulnerability and control, two of the central Stavar features, plainly perform real analytical work in the High Court’s jurisprudence (Stewart & Stuhmcke 2020, p. 173).
Statutory Intervention: The Civil Liability Acts
The common law evolution was then partly overtaken by statute. The collapse of HIH Insurance in 2001 and a sharp escalation in public liability premiums produced a political crisis, and the Commonwealth commissioned the Review of the Law of Negligence chaired by Ipp JA, which recommended a nationally uniform statute restating and limiting negligence liability (Ipp et al. 2002, p. 27). What eventuated was less tidy. Each state and territory enacted its own legislation, including the Civil Liability Act 2002 (NSW), the Civil Liability Act 2003 (Qld) and substantial amendments to the Wrongs Act 1958 (Vic), similar in architecture but divergent in detail (McDonald 2018, p. 41).
For present purposes the critical point is that the Acts largely bypass the duty question. They restate breach through provisions such as section 5B of the New South Wales Act, recalibrate causation, and erect defences around obvious risks and dangerous recreational activities, while leaving the anterior existence of a duty to the common law. The principal exception is mental harm. Part 3 of the New South Wales Act provides that no duty to avoid causing psychiatric injury arises unless the defendant ought to have foreseen that a person of normal fortitude might suffer a recognised psychiatric illness, and section 30 confines recovery for pure mental harm largely to plaintiffs who witnessed the victim being put in peril or who are close family members. These provisions absorb, and partly freeze, the position reached at common law in Tame v New South Wales (2002), as their interpretation in Wicks v State Rail Authority (NSW) (2010), permitting police rescuers at the Waterfall rail disaster to proceed, confirms (Handford 2017, p. 96).
The result is a hybrid and federated structure. Duty is governed by common law, breach by statute, psychiatric injury by a partial codification, and each element varies at the margins between jurisdictions. A claimant injured near the Murray may find that materially different rules apply depending on whether New South Wales or Victorian law governs the claim. The coherence that the High Court demands of common law doctrine is therefore not matched by coherence across the federation, a point pressed repeatedly in the Australian literature (Luntz et al. 2021, p. 152; McDonald 2018, p. 55).
Novel Duties and the Question of Coherence
How, then, does an Australian court approach a novel duty claim today? The practical sequence is reasonably settled. The court first asks whether the case falls within, or closely analogous to, an established category. If it does not, the court undertakes a salient features evaluation of the Stavar kind, in which vulnerability and control usually carry the greatest weight. Any provisional duty is then tested for coherence against the defendant’s statutory functions and against neighbouring doctrines, before the forum’s civil liability provisions are applied. Hunter and New England Local Health District v McKenna (2014) illustrates the coherence check in operation: a hospital that discharged a mentally ill patient who later killed his friend owed no duty to the victim’s relatives, because the Mental Health Act 1990 (NSW) obliged the hospital to prefer the least restrictive form of care, an obligation inconsistent with a common law duty pointing towards continued detention. Similarly, Stuart v Kirkland-Veenstra (2009) held that a statutory power vested in Victoria Police to apprehend a person at risk of suicide did not, without more, generate a common law duty to exercise it.
The standard objection to this settlement is indeterminacy. Seventeen unranked factors do not decide cases; judges must select and weight them, and critics contend that the catalogue functions as a menu from which conclusions reached on undisclosed grounds are retrospectively justified, leaving outcomes difficult to predict and litigation costly (Plunkett 2018, p. 212; Robertson 2019, p. 534). There is force in the objection, but it understates three matters. First, the overwhelming majority of negligence claims fall within established categories where duty is uncontroversial, so the multifactorial method governs only the contested frontier. Secondly, the salient features approach is transparent in a way its predecessors were not: vulnerability, control and indeterminacy are genuine reasons, openly weighed, whereas proximity was a conclusory label that concealed the real basis of decision (Sappideen, Vines & Watson 2016, p. 183). Thirdly, incrementalism is a defensible posture for a final court that has twice watched grand theory fail. On this view the law has not drifted into disorder; it has matured into candour.
The genuinely troubling incoherence is structural rather than methodological. Coherence reasoning operates asymmetrically, defeating duties wherever a statutory scheme occupies the field while seldom grounding new protection, and the civil liability statutes fracture the single national common law that the High Court’s doctrine presupposes. The Ipp panel’s ambition of one national restatement remains unrealised more than two decades after its report (Ipp et al. 2002, p. 29). Harmonisation of the statutory layer, rather than the invention of another judicial formula, is accordingly the reform most likely to improve the coherence of Australian negligence law.
Conclusion
The evolution of the duty of care in Australia describes an arc from aphorism to method. Donoghue v Stevenson supplied the moral generalisation; Grant v Australian Knitting Mills domesticated it; Jaensch v Coffey built proximity upon it; Sullivan v Moody ended the search for a single test and substituted coherence; and Caltex Refineries v Stavar organised what remained into a disciplined, if open ended, evaluation of salient features. The resulting approach is defensible. It confines uncertainty to genuinely novel cases, exposes its real reasons to scrutiny, and respects the limits of judicial generalisation. Its weaknesses, an asymmetric use of coherence that tends only to restrict liability and the fragmentation introduced by eight divergent civil liability regimes, are real, but they are better addressed through legislative harmonisation than through a new unifying formula. Measured against nearly a century of failed tests, the present settlement is not an abandonment of principle. It is the form that principle takes when a mature legal system stops pretending that a single sentence can decide every case.
References
Dietrich, J 2020, ‘Coherence and the duty of care in negligence’, Torts Law Journal, vol. 26, no. 1, pp. 5-27.
Handford, P 2017, ‘Psychiatric injury and the civil liability legislation: unfinished business’, University of Western Australia Law Review, vol. 42, no. 1, pp. 84-108.
Ipp, D, Cane, P, Sheldon, D & Macintosh, I 2002, Review of the Law of Negligence: Final Report, Commonwealth of Australia, Canberra.
Luntz, H, Hambly, D, Burns, K, Dietrich, J, Foster, N, Grant, G & Harder, S 2021, Torts: Cases and Commentary, 9th edn, LexisNexis Butterworths, Sydney.
McDonald, B 2018, ‘The Civil Liability Acts and the fragmentation of Australian negligence law’, Sydney Law Review, vol. 40, no. 1, pp. 39-63.
Plunkett, J 2018, The Duty of Care in Negligence, Hart Publishing, Oxford.
Robertson, A 2019, ‘The salient features approach and the structure of the duty of care inquiry’, Melbourne University Law Review, vol. 43, no. 2, pp. 514-547.
Sappideen, C, Vines, P & Watson, P 2016, Torts: Commentary and Materials, 12th edn, Lawbook Co, Sydney.
Stewart, P & Stuhmcke, A 2020, Australian Principles of Tort Law, 5th edn, Federation Press, Sydney.
Stickley, A 2016, Australian Torts Law, 4th edn, LexisNexis Butterworths, Chatswood.