When a GLO Test Case Is No Longer the Last Word: Supreme Court Guidance in AXA v HMRC

When does a GLO test case stop being the last word? That was the central question for the Supreme Court in AXA Insurance UK PLC and another v Commissioners of Inland Revenue and another [2026] UKSC 24, which considered when a judgment in a group litigation test case should not bind follower claims after the law has subsequently changed. The judgment goes to an important aspect of the Group Litigation Order (GLO) regime: balancing finality and consistency across large cohorts of claims against fairness to parties who would be impacted by not having their cases decided in line with more recent legal developments. 

Background

GLOs allow common or related issues across multiple claims to be managed together. A key feature of the regime is that a judgment on a GLO issue in one claim generally binds the parties to all other claims on the group register, unless the court “orders otherwise” under CPR r 19.23(1)(a). The Supreme Court’s decision in this case concerned the scope of that exception.

The GLO at issue was the Controlled Foreign Companies and Dividend Group, formed in July 2003 and involving claims against UK tax authorities, including the Revenue. The claims centred on the allegation that the tax treatment of dividends received from non-resident companies was less favourable than that of dividends from UK-resident companies, and therefore incompatible with EU law.

Three test cases were chosen to go forward, including the “Prudential test case”. The Supreme Court decision, however, concerned the stayed “AXA proceedings”, which were follower claims rather than test cases, and which raised two main issues: entitlement to interest and limitation.

Procedural history

The Prudential test case was decided by the High Court in 2013. It was later appealed to the Court of Appeal in 2016, and the Supreme Court in 2018. When the Prudential test case was decided, the stay imposed on the AXA proceedings was lifted.

The taxpayers in the AXA proceedings, seeking to rely on the binding effect of the Prudential test case, being a GLO test case judgment, submitted that they could rely in the follower claims on the issues that were raised and decided in the first instance Prudential test case. However, the state of the law at the time of that first instance judgment then underwent significant changes, which included the Supreme Court overturning the judgment in Sempra Metals [2007] UKHL 34 in Prudential [2018] UKSC 39 on the question of interest. The Revenue argued in the AXA proceedings that the relevant parts of the first instance Prudential decision should no longer bind the follower claims because they rested on a legal basis now wrong.

The High Court decided that the outstanding issues in the AXA proceedings had not been decided in the Prudential case and should therefore instead be determined in accordance with the law as it currently stood. The Court of Appeal upheld this decision, save that it held that one of the outstanding issues had been decided as a GLO issue in the first instance Prudential judgment. It therefore had to decide whether to “order otherwise” under CPR r 19.23. It held that there was a strong reason to do so: it would be wrong to determine the follower claims on a basis now known to be incorrect in law. The appellants then appealed to the Supreme Court.

The Supreme Court decision

The Supreme Court upheld the Court of Appeal’s decision. It confirmed that the power to “order otherwise” under CPR r 19.23 is narrow and the ordinary rule is that a judgment on a GLO issue binds other claims on the group register. However, that rule is not absolute. Drawing an analogy with estoppel cases, where the Supreme Court accepted that, notwithstanding the narrow scope of CPR r 19.23, finality and consistency may give way where it is clearly established that the legal basis of the earlier decision was wrong and applying it would cause injustice. It is not enough that a party now wishes to raise different or better arguments.

On the interest issue, the Court held that finality did not justify applying an erroneous legal rule to the AXA proceedings. To do so would impose liability on the Revenue on a basis now understood to be wrong in law.

The limitation issue was different: the Prudential proceedings had not determined the limitation-start date for every claim in the group. AXA’s limitation position therefore had to be assessed by reference to its own circumstances, rather than treated as automatically resolved by the test case.

The Court put the point starkly: if the claims were determined today outside the GLO context, the relevant submissions – on either interest or limitation – would fail. The case therefore fell within the limited category where the Court may “order otherwise” to prevent the binding effect of a test case judgment.

Comment

The decision is a useful reminder that CPR r 19.23(1)(a) is not a broad licence to depart from GLO test cases. The default position remains finality and consistency across the GLO. But where the legal foundation of a binding test case judgment has clearly fallen away, it is now established that the court may intervene to avoid injustice. Parties to GLO cases can be reassured that test cases will still bind in the ordinary course of events, but it is nonetheless important to be aware that their binding effect is not absolute, and the impact of appeals must be borne in mind.