IR35 Mutuality of Obligation & Control: The Real Battleground

Behind almost every IR35 dispute sits the same foundational question the tribunals keep returning to: mutuality of obligation and control. HMRC's CEST tool treats these lightly, but a decade-long piece of litigation involving football referees, which finally concluded in 2026, shows just how much genuine argument still lives in this part of the test, and why it so often decides the outcome of a case.

The three-stage test everything else sits on top of

Employment status for tax purposes still starts with the test set out by MacKenna J in Ready Mixed Concrete (South East) Ltd v Minister of Pensions and National Insurance [1968] 2 QB 497, since described by the Supreme Court as the “classic description of a contract of employment”. A contract of service exists where the individual agrees to provide their own work and skill in return for pay, agrees to be subject to the engager's control to a sufficient degree, and the other terms of the arrangement are consistent with employment. The first two conditions, mutuality of obligation and personal service form the first stage, and control forms the second, are necessary but not sufficient. Only if both are present does the tribunal move to the third stage: a multi-factorial assessment of all other relevant factors, financial risk, integration, provision of equipment and the wider picture, to decide whether the relationship really is one of employment.

The Court of Appeal in HMRC v Atholl House Productions Ltd [2022] EWCA Civ 501 confirmed there is no real conflict between this three-stage structure and the broader “all relevant factors” approach seen in other cases: the multi-factorial assessment is simply what the third stage requires. In practice, most IR35 arguments concentrate heavily on stages one and two, mutuality and control, because they are treated as gateway conditions, and if either is missing, employment status is out of reach regardless of how the third-stage factors might otherwise point.

Mutuality of obligation: narrower than it sounds

Mutuality of obligation simply means the individual agrees to do the work and the engager agrees to pay for it. That mutual exchange exists in almost every tax dispute, because tax cases are about payments actually made for work actually done, not about whether there was an ongoing obligation to offer and accept future work during gaps between assignments. Genuine absence of mutuality tends to arise in different circumstances: where the individual has no obligation to turn up and the engager has no obligation to offer work if available (Carmichael and Another v National Power [1999] 4 All ER 897), or where the individual works but is paid by third parties rather than the engager, as with the lap dancer in Quashie v Stringfellows Restaurants Limited [2012] EWCA Civ 1735. The Court of Appeal in Secretary of State for Justice v Windle & Arada [2016] EWCA Civ 459 went further, holding that genuinely casual, intermittent working can itself support an inference of self-employment.

A recurring misconception is that a contract terminable at will, by either side, without notice or breach, defeats mutuality. It does not. The Court of Appeal in PGMOL v HMRC [2021] EWCA Civ 1370 held that where a contract exists, the fact its terms allow either party to end it before performance, without breaching it, is immaterial, the contract subsists with its mutual obligations for as long as it runs, and the Supreme Court upheld that reasoning in 2024. This closed off what had been a common contractor argument: that a right to walk away, or a client's right to end an engagement without notice, by itself defeated mutuality. It generally does not.

Control: wider than a right to instruct

Control requires the individual to be subject to the engager's authority “in a sufficient degree to make that other master”. The Supreme Court in Professional Game Match Officials Ltd v HMRC [2024] UKSC 29 described this as allowing for a wide range of circumstances, to be assessed on the facts of each case, and confirmed that satisfying the control condition does not require a contractual right to intervene in every aspect of how the work is performed. Control “may take many forms” and is not confined to direct instructions, but “there must be some control, if only in incidental or collateral matters”. The Supreme Court also confirmed that the availability of sanctions, the ability to remove someone from future work rather than direct them in real time, is itself a means of exercising control, control can be exercised by positive or negative means alike.

The Ready Mixed Concrete judgment itself divides control into control as to how, when, where and what work is done, and tribunals routinely work through each of these separately rather than treating control as a single yes/no question. In Atholl House and related cases, the Upper Tribunal rejected the argument that control over precisely what work is done is always required for a sufficient framework of control to exist, in appropriate cases, a looser but still real framework of control over when and how the work happens can be enough.

The PGMOL saga: what a decade of litigation actually settled

The full history of PGMOL v HMRC illustrates both how contested this area is and how little, ultimately, mutuality and control alone decide. HMRC assessed Professional Game Match Officials Ltd on the basis that Level 1 football referees, engaged match by match as self-employed officials, should have been treated as employees. The First-tier Tribunal and Upper Tribunal both found insufficient mutuality and control. The Court of Appeal overturned that in 2021, holding that mutuality persisted for the duration of each individual match engagement regardless of the parties' ability to cancel beforehand. The Supreme Court substantially agreed in September 2024, confirming that mutuality and control could exist within these short, engagement-by-engagement arrangements, and remitted the case to the tribunal to complete the third-stage analysis with that clarified understanding of stages one and two.

The First-tier Tribunal's final decision, reported in 2026 as PGMOL v HMRC [2026] UKFTT 654 (TC), found that although mutuality of obligation and control were indeed present within each individual match engagement, the overall relationship was still not one of employment once the third-stage factors, including the lack of any obligation between matches, the referees' other work, and the informal, non-integrated nature of the wider relationship, were weighed. HMRC did not appeal the final result. The case stands as the clearest illustration available that winning the mutuality and control argument, which HMRC did, does not win the case; it only earns entry to the third stage, where the wider picture still has to support an employment finding.

Why this matters beyond referees: the umbrella and gaps-between-assignments point

The mutuality analysis has direct relevance to umbrella companies and contractors working intermittent assignments through the same intermediary. In Mainpay Ltd v HMRC [2025] EWCA Civ 1290, upholding the First-tier and Upper Tribunal, the Court of Appeal confirmed that mutuality was lacking in the gaps between assignments for workers supplied by an umbrella company, meaning there was no single overarching contract of employment spanning multiple placements, each assignment instead constituted its own separate employment. The practical consequence in that case was that travel to each assignment counted as travel to a permanent workplace rather than a temporary one, affecting the tax treatment of travel expense claims under sections 338 and 339 of ITEPA 2003. For contractors and umbrella arrangements more generally, whether an overarching contract exists across gaps between engagements remains a genuinely live, fact-specific question distinct from the mutuality that exists within any single engagement while it runs.

How these cases typically unfold

Case A: The contractor who thought a break clause settled it

A contractor engaged on a rolling series of one-month IT support contracts argues that because either party could terminate on a week's notice without cause, there was never any real obligation on the client to keep offering work or on the contractor to keep accepting it, and therefore no mutuality. HMRC, citing the Court of Appeal and Supreme Court reasoning in PGMOL, points out that during each live one-month contract, both parties were bound by mutual obligations, work was actually done and paid for, and the termination right does not retrospectively undo that. The mutuality argument fails, and the case turns instead on the genuinely contested third-stage factors, financial risk, whether the contractor used their own equipment, and the degree of integration into the client's team.

Case B: The genuine substitution right that succeeded

A contractor's agreement includes a substitution clause the client accepted at face value: on two occasions during the engagement, the contractor sent a suitably qualified colleague in their place, invoiced by the contractor's own company, with the client raising no objection and paying without requiring any approval process. HMRC challenges the clause as theoretical, but the documented evidence of genuine substitution actually happening, unprompted and without client veto, is enough to establish that personal service, a separate necessary condition alongside mutuality, was not present, and the IR35 status is successfully defended on that basis without the case needing to reach a detailed control or third-stage analysis at all.

What this means for status reviews going forward

Building an IR35 or off-payroll status position solely around mutuality or a contractual termination clause is now a significantly weaker strategy than it was before PGMOL reached the Supreme Court. Genuine substitution rights, actually capable of being exercised and evidenced if they are, remain one of the stronger levers available, because personal service sits alongside, rather than within, the mutuality and control analysis. Beyond that, status arguments increasingly have to be won, or lost, on the wider third-stage picture, financial risk, integration, equipment and the overall character of the relationship, which is exactly where HMRC's CEST tool is weakest and where a properly reasoned, evidence-based Status Determination Statement earns its value.

Related guides in this series

• IR35 investigations: the full picture
• HMRC’s CEST tool: what it gets wrong
• Defending an IR35 enquiry
• Status Determination Statements explained
• Managed Service Company legislation explained
• Kickabout Productions v HMRC [2022] EWCA Civ 502