The Supreme Court resets the law on deprivation of liberty

A Reference by the Attorney General for Northern Ireland [2026] UKSC 16

For twelve years, the practical question “is this person deprived of their liberty?” has had a deceptively simple answer. Since Surrey County Council v P; Cheshire West and Chester Council v P [2014] UKSC 19; [2014] AC 896 in 2014, if a person lacked the mental capacity to consent to their care arrangements, was under continuous supervision and control, and was not free to leave, they were deprived of their liberty – and the State had to authorise and review that deprivation. That was the famous “acid test”, and Baroness Hale’s phrase “a gilded cage is still a cage” became its emblem.

On 2 June 2026, a seven-Justice panel of the Supreme Court unanimously overruled that decision. The acid test no longer governs the question on its own. In its place is a return to the older, more open-textured “multifactorial” assessment that the Strasbourg court has always applied. And – just as importantly – the Court has revived a second element of the test that Cheshire West had effectively switched off: the question of whether the person has actually, in the Article 5 sense, given valid consent to the confinement and care arrangements that make up their concrete situation.

This is not a narrow Northern Irish ruling, even though that is how it arrived at the Court. The mental capacity legislation in England and Wales (the MCA 2005) and in Northern Ireland (the MCA 2016) ties the meaning of “deprivation of liberty” directly to Article 5 of the European Convention on Human Rights. Scotland reaches the same destination by a different road: the implications there arise from the Convention-based limits on the competence of the Scottish Parliament and the Scottish Ministers under the Scotland Act 1998, and from the Human Rights Act 1998, rather than from an identical definitional link in Scottish mental capacity legislation. Either way, the decision reshapes the law across all four nations.

How the case arrived: a reference about a Code of Practice

This was an unusual vehicle for such a seismic decision. There was no individual claimant and no concrete set of facts. Instead, the Attorney General for Northern Ireland referred a question of law to the Court under the devolution provisions of the Northern Ireland Act 1998.

The background was practical. The Minister of Health for Northern Ireland wanted to change practice so that a person who lacks capacity to decide on their care arrangements could nonetheless give valid consent to those arrangements by expressing wishes and feelings that go beyond mere acquiescence. The question referred was whether the Minister of Health for Northern Ireland has the power to revise the Deprivation of Liberty Safeguards Code of Practice so that persons aged 16 and over who lack capacity to make decisions about their care and treatment can give valid consent to their confinement through the expression of their wishes and feelings. The draft “Revised Code” is reproduced at Annex 1 of the judgment.

Two features of this procedural posture deserve emphasis, because they shaped what the Court could and could not do.

(a) A prospective (“ab ante”) review. The Court was asked to rule on the lawfulness of guidance before it was issued and applied to anyone. The judgment is candid that this is an awkward exercise (paras 35–40). The Court also doubted whether a Code of Practice is really “subordinate legislation” at all – noting that a code is guidance, “a statement of general policy”, rather than legislation in the ordinary sense (para 17) – but held nothing turned on the point.

(b) The applicable test. Because this was guidance rather than primary legislation, the Court did not apply the highly deferential “incompatible in all or almost all cases” standard used for devolved legislation. Instead it applied the test from R (A) v Secretary of State for the Home Department [2021] UKSC 37: does the policy “authorise or approve unlawful conduct” in a significant number of cases (paras 39 – 40)? This matters: the Court was effectively asking whether the Revised Code would mislead decision-makers into breaching Article 5.

There is a logical pivot here that the whole judgment turns on (paras 40 – 41): if the acid test in Cheshire West was good law, the Revised Code would be incompatible with it and therefore unlawful. So the Court could not answer the reference without confronting whether Cheshire West itself was correctly decided.

The constitutional headline: a rare use of the 1966 Practice Statement

The Supreme Court does not lightly overrule itself. The power to do so derives from the House of Lords’ 1966 Practice Statement ([1966] 3 All ER 77), and the Court reaffirmed that it is exercised “rarely and sparingly”, recalling Lord Bingham’s warning in Horton v Sadler [2006] UKHL 27; [2007] 1 AC 307 that it has never been enough that a later generation of judges would simply have decided the matter differently (para 44).

So why depart here? The Court gave several interlocking reasons (paras 46–51):

• The definitional link to Article 5. Because the mental capacity statutes in England and Wales and in Northern Ireland define deprivation of liberty by reference to Article 5, any gap between the domestic interpretation and Strasbourg’s creates real, practical confusion. The Court considered that it has a particular responsibility to ensure that the interpretation of the Convention rights in Article 5 is properly in line with the Strasbourg jurisprudence.
• No corrective route via Strasbourg. Public authorities cannot themselves take the UK to Strasbourg to fix an over-broad domestic reading, so the error could fester indefinitely (para 49).
• The administrative consequences. Following R v National Insurance Comr, ex p Hudson [1972] AC 944, 1024 (Lord Simon of Glaisdale), the Court treated the administrative difficulties generated by the decision as a further ground for departure (paras 50, 99–113).

One subtlety worth flagging for advanced readers: the Court was careful to note that there had been no relevant change in the Strasbourg case law since 2014 (para 47). This was therefore not the easy case where a later Strasbourg ruling supersedes a domestic one. The Court had to confront, head-on, whether the Cheshire West majority had simply misread the Strasbourg material as it stood – and concluded that it had.

What Cheshire West actually decided – and the flaw the Court identified

To understand the change, recall the structure of the Article 5 test as laid down by Strasbourg in Storck v Germany (2005) 43 EHRR 6. A deprivation of liberty has three elements (para 4):

1. Objective element – confinement in a restricted space for a non-negligible time;
2. Subjective element – the absence of valid consent to that confinement;
3. Attribution – responsibility of the State.

In Cheshire West, the parties treated only the objective element as live. It appears to have been common ground, and never argued, that because the three individuals lacked capacity under the Mental Capacity Act, they could not give valid consent – so the subjective element was simply assumed to be satisfied (paras 82 – 84).

The 2026 Court identified this as the original sin of the case. It described the position as “both unfortunate and confused” (para 84). The reason is that valid consent under Article 5 is an autonomous Convention concept – it is governed by Strasbourg principles, not by domestic capacity law. It therefore does not follow that a person who lacks legal capacity under the Mental Capacity Act is unable to give valid consent for Article 5 purposes (para 84).

Because the majority assumed the subjective element away and focused only on the objective element, the acid test they formulated – “continuous supervision and control” and “not free to leave” – came to govern the whole of Article 5, including its subjective component (para 97). The 2026 Court held that the majority had thereby:

• equated lack of legal capacity with lack of valid consent, wrongly collapsing two distinct concepts (para 201);
• treated the person’s compliance, lack of objection, the relative normality of the placement, and the purpose of the confinement as all irrelevant (paras 86, 89); and
• in doing so, departed from the long-standing multifactorial approach (para 183).

A point that may surprise practitioners: the Court even suggested the majority got a key piece of analysis “the wrong way round” (para 85). Baroness Hale had appeared to think that a person incapable of expressing a view was thereby deprived of liberty; on the Court’s reading of Strasbourg, the inability to express any view at all tends to point away from a deprivation (more on the catatonic and unconscious cases below).

The new framework: what the law now is

The Court’s conclusions are summarised at paras 52 – 55. Stripped to essentials, the law is now as follows.

A multifactorial assessment, not a bright line

The starting point is the individual’s “concrete situation”, assessed by reference to a whole range of factors – the type, duration, effects and manner of implementation of the measures in question. No single factor is determinative (para 53(i)).

Crucially, the Court did not say the acid test is meaningless. It said it is not sufficient on its own. In a passage practitioners will return to repeatedly, the Court put it cautiously: it “may well be” that a person cannot be deprived of liberty without the acid test being met, but meeting it is not enough (para 184). On that deliberately tentative formulation, satisfying the acid test opens the inquiry rather than closing it; the Court stopped short of holding definitively that the acid test is an indispensable threshold in every case.

The revived subjective element: valid consent without capacity

This is the doctrinal heart of the case. The Court held that a person who lacks legal capacity may nonetheless give valid consent to their confinement for Article 5 purposes. The principle is set out at para 53(ii): an individual who is conscious of their environment, has a basic understanding of their circumstances, and can express a view, and who manifests acceptance of their situation, should have that opinion respected.

It is important to be precise about what is being consented to. It is not simply whether the person likes their accommodation. It is whether they validly consent, in the Article 5 sense, to the confinement and the living and care arrangements that make up their concrete situation. The Court drew a careful and important distinction between two quite different ideas of “consent” (para 126):

• Valid consent in the Article 5 sense – an aspect of whether there is a deprivation at all. This turns on the person’s de facto understanding and lived experience.
• Consent as a waiver of rights – which requires free, fully informed consent and legal capacity. This is a different doctrine and is not what is in play here.

This distinction matters because it answers the obvious objection – “how can someone who lacks capacity consent to anything?” The answer is that they are not waiving a right; they are simply telling us, in a basic human way, whether they are content with the situation in which they are being kept and cared for. As the Court put it, whether someone experiences their situation as detention against their wishes or as a supportive home where they are content “is a basic aspect of human experience”, and there is no good reason to disregard the individual’s own evaluation of it (para 135).

The Court grounded this in a close reading of the Strasbourg authorities. The load-bearing authorities are the Grand Chamber decision in Stanev v Bulgaria (2012) 55 EHRR 22, GC, and H.M. v. Switzerland, no. 39187/98, § 42, ECHR 2002-II. Mihailovs v Latvia [2014] MHLR 87 supplies the cleanest illustration of the key point – that tacit acceptance of an open placement the person could object to negates a deprivation (the Lielberze home: para 172). A recurring proposition – that lack of legal capacity does not entail an inability to understand one’s situation in fact (the question being the person’s factual ability to express a view: paras 123, 164) – is traced through Shtukaturov v Russia (2012) 54 EHRR 27, para 108, as picked up in turn by DD v Lithuania [2012] MHLR 209, Kędzior v Poland [2013] MHLR 115 and Mihailovs v Latvia [2014] MHLR 87. The Court also took care to read Stankov v Bulgaria (Application No 25820/07) (unreported), judgment of 17 March 2015, narrowly, warning that para 90 of that judgment ran the deprivation-of-liberty question together with a different test (loss of victim status, drawing on M v Ukraine [2013] MHLR 255, para 77) and must not be over-read (paras 177–179).

The safeguard against over-reach is built in: where there is serious doubt, no inference of consent should be drawn (para 53(v), para 191).

5.3 Physical liberty and the “catatonic” cases

The Court re-emphasised that Article 5 protects physical liberty, not autonomy in the abstract (para 118). This generated one of the judgment’s more striking conclusions. A person who, by reason of their condition, has no physical liberty to exercise at all – someone catatonic, in a minimally conscious state, or so profoundly disabled that they cannot conceptualise leaving, let alone physically achieve it – is not being deprived of liberty (paras 142, 196–199). As the Court put it, drawing on Professor Eldergill, “mere inability is not lack of freedom” (para 197). The position of a temporarily unconscious person is different and depends on context, including their realistic potential to regain consciousness and whether, on doing so, they would be prevented from leaving (paras 143–145): a detained patient or prisoner remains deprived of liberty while unconscious, whereas ordinary medical treatment usually will not amount to a deprivation.

It is important not to over-read this. The Court was addressing the person who genuinely cannot form or act on the idea of leaving. It expressly distinguished – and reserved – the different case of a person who is mentally alert but physically unable to leave: the State may in some circumstances owe positive obligations, for example to provide a wheelchair or other aid to enable the individual to leave, but the Court said that was “a wholly different point” which it did not decide (para 198). So this strand of the judgment narrows Article 5 for the profoundly cognitively impaired, but it does not resolve the position of the physically disabled but cognitively able.

The Court was alert to the equality objection – does this give disabled people a lesser right? It said no. Liberty means the same for everyone; but a person who cannot leave is in a fundamentally different position from one who is prevented from leaving, so there is no less favourable treatment of people in a comparable situation (paras 198 – 199).

Coercion, normality and purpose are back in play

Three factors Cheshire West declared irrelevant are now relevant again:

• Compliance / absence of objection bears on whether there is confinement at all, because the effect of arrangements differs depending on whether the person is content or resisting (para 187). Coercion is described as close to a necessary feature: if a person can and does express contentment, “it will ordinarily be difficult to see how they are being coerced” (para 189). Mere passive acquiescence, especially where sedating medication may be suppressing objection, counts for little (para 188).
• Relative normality of the setting matters – the Court noted that no Strasbourg decision has ever held that a person living in their own home is deprived of liberty, though it stopped short of saying it could never happen (para 193).
• Purpose of the measure is a relevant (though not decisive) factor in borderline cases (paras 130 – 134, 200).

5.5 The fate of MIG, MEG and P

In a remarkable epilogue, the Court re-examined the three individuals from Cheshire West itself. It held that MIG and MEG were not deprived of their liberty – there were sufficient indications they were content, their arrangements were close to normal, and the objective element was not made out (paras 203–204). As to P, the Court declined to overrule that specific finding (it had not heard argument on his facts) but expressed real doubt, indicating sympathy with Munby LJ’s view in the Court of Appeal that P was not deprived of his liberty (para 205).

The answer to the reference

Having dismantled the acid test, the answer to the reference followed straightforwardly. Lord Sales and Lady Simler said they would decline to follow Cheshire West and would overrule it; the so-called acid test has never been adopted by the European court and is wrong in principle. The Minister would not act incompatibly with Article 5 by issuing the Revised Code, and so it is within his competence to do so (paras 207 – 208).

But two heavy caveats sit on that conclusion. First, the Court noted the draft Revised Code “has not been fully worked through and in any event, it will require revision in the light of this judgment” (para 209). Second – and this is easy to miss – my own reading is that the Court was distinctly guarded about the Code’s central drafting device, the “positive attitude to the care arrangements” test in para 2.12 of the Code. I should be clear that this is analysis, not a holding: the Court did not reject that formulation, and indeed held the Minister would be acting within competence. What it did was acknowledge the real difficulty of telling whether a profoundly disabled person is agreeing, objecting, or simply presenting (para 55), and warn that this “will require careful consideration in finalising the Revised Code”. The fair reading, in my view, is that the judgment endorses the principle (capacity ≠ consent) more confidently than it endorses the Code’s particular formulation of it – but readers should weigh that as my inference.

How the new approach compares to Cheshire West

Issue	Cheshire West (2014)	AG for NI (2026)
Test for deprivation	“Acid test”: continuous supervision and control + not free to leave	Multifactorial assessment of the concrete situation; acid test not sufficient on its own
Status of the acid test	Effectively determinative	“May well be” a precondition, but the Court did not decide this definitively; in any event not sufficient
Lack of capacity and consent	Treated as equivalent to lack of valid consent	Distinct; a person without capacity may still give valid consent in fact
Compliance / absence of objection	Irrelevant	Relevant to whether there is confinement
Relative normality of placement	Irrelevant	Relevant
Purpose of the measure	Irrelevant	Relevant (not decisive) in borderline cases
Person with no physical liberty to exercise (catatonic, minimally conscious, profoundly impaired)	Treated as deprived (“err on the side of caution”)	Not deprived – no physical liberty to exercise (temporary unconsciousness treated separately; positive-obligation cases reserved)
Guiding metaphor	“A gilded cage is still a cage”	Proximity to “the paradigm of confinement in a prison cell”

Practical consequences

The numbers problem the Court was trying to solve

A substantial part of the judgment (Section 7) is devoted to the administrative aftermath of Cheshire West, and it is essential context for understanding why the Court acted. The decision caused authorisation applications to rise roughly tenfold (paras 99 – 101), generating enormous backlogs; the Law Commission had estimated full compliance would cost around £2.2 billion per year (para 104). The Court described the “paradoxical outcome” that a person positively happy with their living arrangements is nonetheless classed as deprived of liberty and subjected to intrusive assessments (para 109). It was plainly troubled that scarce resources were being diverted from those who genuinely need urgent review (para 103).

The likely practical effect is a substantial fall in the number of cases requiring authorisation, though the precise scale is uncertain. The judgment records that over 400,000 people across the United Kingdom are currently treated as deprived of liberty (para 13); it does not suggest that all, or even most, of them were content with their arrangements. What it does identify is a system that swept up very large numbers of people, including many who were positively happy with where they lived and had no wish to leave (paras 20, 108–109). Not all of those people will now fall outside Article 5 – but many may.

The assessment task gets harder, not easier

There is an irony here that every front-line practitioner should grasp. Cheshire West was administratively burdensome but conceptually easy: the acid test was, in the charities’ words to the Court, clear and workable (para 28). The new approach is administratively lighter but conceptually harder. The “deprivation of liberty: yes/no” question – which under Cheshire West was rarely disputed – now reopens in every case, requiring a holistic, fact-sensitive judgment about supervision, restriction, normality, purpose, and the person’s own wishes and feelings.

The Court was realistic that eliciting wishes and feelings from people with profound impairments, fluctuating presentations, or a tendency to acquiesce or mask will be genuinely difficult – these were powerful points in the charities’ evidence (paras 30, 191). Its answer is the “serious doubt” rule: if in doubt, draw no inference of consent (para 191). In practice, that rule will do an enormous amount of work, and how conservatively it is applied will determine whether this judgment narrows Article 5 modestly or dramatically.

The access-to-justice concern – the elephant in the room

The most serious objection, pressed by the charities and the Official Solicitor, is not really answered to everyone’s satisfaction. The logic is stark: if a person is, objectively, subject to confinement, but is held not to be deprived of liberty for subjective reasons, the restrictions on them do not change one inch – what changes is that they lose the Article 5 safeguards altogether. That loss is not merely procedural. They lose the substantive guarantee under Article 5(1)(e) that detention on grounds of unsound mind is lawful only if it is objectively justified – broadly, that there is a true mental disorder, of a kind or degree warranting confinement, which persists throughout. And they lose the procedural protections that go with it: independent advocacy and the right under Article 5(4) to go to court to test their detention (para 31).

The Court’s response operates on two levels. First, doctrinally: if there is no deprivation of liberty, Article 5(4) is simply not engaged – that is how the concept works. Second, and more substantively, the Court repeatedly stressed that these individuals are not left unprotected. They retain rights under Articles 2, 3 and 8, the protections of best-interests decision-making under the mental capacity statutes, local authority safeguarding duties, and common law duties of care (paras 10, 141, 185). The Court considered Baroness Hale’s contrary “err on the side of caution” rationale to be “overstated” precisely because of these other protections (para 141).

Whether that reassurance holds in practice is the central question this judgment leaves open. Article 8 and safeguarding processes do not deliver automatic, independent, periodic review with access to a court and legal aid in the way Article 5 does. For the person whose objective confinement is real but whose “consent” is inferred from a thin evidential base, the practical position is that the Article 5 route to challenge – including Article 5(4) and the DoLS/authorisation review machinery – will not be engaged at all. That does not mean such a person is without recourse: the Court of Protection’s welfare jurisdiction does not disappear merely because Article 5 is not in play, and a public law challenge may be available depending on the issue. But those routes would have to do more of the work that Article 5 currently does, and they are not obvious like-for-like substitutes.

The effect of the overruling, and existing authorisations

The judgment takes effect in the ordinary way of an overruling: as a declaratory statement of what the law always was, it means the Cheshire West approach should no longer be applied, and the many guidance documents and codes that embed the acid test now have to be read in that light. I should flag that the Court gave no express transitional direction and set no grace period – the points that follow are therefore practical and legal inference rather than something the Court spelled out. Existing authorisations and orders do not automatically become unlawful simply because the underlying test has changed; but a great many cases will, in due course, need to be revisited to ask the new, prior question: is this person deprived of liberty at all? The mechanics of unwinding live authorisations are genuinely complex and will need to be worked through in guidance.

Wide applicability

Although the reference came from Northern Ireland, the reasoning is UK-wide. The direct definitional link to Article 5 exists in the MCA 2005 (England and Wales) and the MCA 2016 (Northern Ireland). Scotland is affected by a different mechanism: the Convention-based limits on devolved competence under the Scotland Act 1998 and the operation of the Human Rights Act, rather than an identical statutory definition (para 22). The interveners reflected this UK-wide reach: the Lord Advocate, the Counsel General for Wales, the Secretary of State, and the Mental Welfare Commission for Scotland all participated. England and Wales’s stalled Liberty Protection Safeguards regime – never brought into force (para 113) – will now have to be reconsidered in the light of this decision. Annex 2’s comparative table will be a useful starting point for anyone advising across borders.

Critical reflections for practitioners

A few observations that go beyond the headlines:

(1) The judgment is more nuanced than “the acid test is dead.” The acid test survives, at least, as a strong indicator and very possibly as a precondition – though the Court framed even that cautiously and did not decide it conclusively (para 184). The real change is the addition of a revived subjective element and the re-admission of normality, purpose and compliance as relevant factors. Advisers who treat the case as simply abolishing the test will mis-state it.

(2) “Valid consent” will be the new battleground. Expect litigation to migrate from “are the objective conditions met?” to “has this person, in fact, validly consented?” The Court has given the framework but not a formula. The “serious doubt → no inference” rule, the relevance of sedation, and the spectrum of communication abilities (para 191) will be argued case by case.

(3) The reasoning vindicates the instinct behind the lower-court “rebellions” – though not every step of it. The Court treated the repeated attempts of experienced judges to distinguish Cheshire West as evidence that the decision had not provided appropriate guidance and was impeding the proper development of the law (para 117). It specifically endorsed both the conclusion and the reasoning of Lieven J in Re SM (Deprivation of Liberty; Severely Disabled Child) [2024] EWHC 493 (Fam); [2024] 2 FLR 896 (“SM”)  (paras 196, 199). But it did not approve every route taken: it was, for instance, expressly critical of the state-control distinction Arden LJ drew in R (Ferreira) v Inner South London Senior Coroner [2017] EWCA Civ 31; [2018] QB 487, calling it “not persuasive” (para 116). So the fair summary is that the Court validated the lower courts’ discomfort, and some of their conclusions, rather than endorsing all of their reasoning.

(4) The constitutional precedent is significant in its own right. This is a rare, deliberate use of the 1966 Practice Statement to correct what the Court saw as a domestic misreading of Strasbourg – not a response to a later Strasbourg ruling (para 47). It is a notable assertion of the Supreme Court’s role as the guardian of Convention interpretation in domestic law where Strasbourg cannot itself intervene (para 49).

(5) The human stakes are not resolved by the judgment. The Court has narrowed the legal category of “deprivation of liberty”; it has not addressed how, in practice, the State will ensure that those who fall outside Article 5 but remain objectively confined receive meaningful, independent oversight. That gap – between the Court’s confidence in Articles 2, 3 and 8 and the fear of safeguards lost behind closed doors – is where the next decade of policy and litigation will be fought.

10. Conclusion

AG for Northern Ireland is the most consequential mental capacity decision since Cheshire West itself, and a rare instance of the Supreme Court overruling its own recent authority. Its doctrinal core is elegant: it disentangles the Strasbourg concept of valid consent from the domestic concept of capacity, and restores the multifactorial assessment that the European court has always used. Its practical promise is relief from a vast and costly authorisation system that swept up very large numbers of people, including many who were content with their arrangements.

But the judgment buys conceptual coherence and administrative relief at a price that is, as yet, unquantified. By making it harder to establish a deprivation of liberty, it necessarily makes it harder for some objectively confined people to access the independent review that Article 5 uniquely guarantees. The Court’s faith that Articles 2, 3 and 8, best-interests processes and safeguarding duties will fill the gap is a proposition that will be tested, repeatedly, in the years ahead.

For now, the practical message is clear. The first question in every case is no longer “do the arrangements meet the acid test?” It is the older, harder, more human question the Court has restored to the centre of the law: looking at this person’s whole situation, and giving real weight to their own wishes and feelings – is this confinement, against their will, at all?

Articles and content on this site are for informational purposes only and do not constitute legal advice. Do not rely solely on this information.