Tooth v HMRC and the SDLT 20-year discovery window

When HMRC opens a Stamp Duty Land Tax (SDLT) compliance check more than four years after the transaction completed, the entire investigation hangs on one statutory word: deliberate. Under paragraph 30 of Schedule 10 to the Finance Act 2003, the standard discovery-assessment window is four years; carelessness stretches it to six; and a deliberate inaccuracy stretches it to twenty.

The Supreme Court's 2021 ruling in Tooth v HMRC [2021] UKSC 17 narrowed what HMRC must establish to reach for the 20-year window — and it is a higher bar than the previous tribunal authorities had suggested. For SDLT taxpayers receiving a discovery assessment dated more than four years after completion, the ratio of Tooth is the central question.

This article walks through the statutory framework, the Tooth facts, the Supreme Court's reasoning, and how the ruling fits with Hicks v HMRC (on what counts as a "discovery") and the wider Schedule 10 enquiry regime. It is a case-law summary, not advice — speak to a qualified adviser before responding to any HMRC correspondence.

The discovery windows in Schedule 10

Schedule 10 to FA 2003 governs SDLT enquiries, amendments, discovery assessments, and appeals. Discovery sits in paragraphs 28–32. The structure is:

A discovery assessment outside the four-year window is only valid if HMRC can demonstrate the relevant behaviour standard. The 20-year window in particular is reserved for the most serious cases — deliberately filing a return known to contain an inaccurate statement.

For broader context on how HMRC opens an SDLT enquiry within the in-window period, see our SDLT enquiry windows and HMRC compliance check letters explainer.

The pre-Tooth uncertainty

Before Tooth, the leading tribunal authority was the Upper Tribunal's decision in Auld v HMRC, which applied a relatively wide reading of "deliberate inaccuracy" — capturing not only intentional misstatements but also "blind-eye" knowledge: situations where the taxpayer suspected the return was inaccurate and chose not to investigate. That wider reading made the 20-year window more readily available to HMRC.

The Court of Appeal in Tooth went the other way: there was no deliberate inaccuracy because, on its reading of the return as a whole, there was no inaccuracy at all. Both sides appealed to the Supreme Court.

The Tooth facts

The return in dispute was Mr Raymond Tooth's 2007/08 self-assessment income tax return. Mr Tooth had entered into a partnership-based employment loss scheme. His accountants used commercial filing software that did not have a dedicated box for the employment-related loss the scheme generated. To make the figures flow into the right place on the calculation, they entered the loss in the "partnership share of profits" box — and added a "white space" disclosure elsewhere in the return explaining exactly what they had done and why.

HMRC opened the discovery assessment in October 2014 — more than four years after the return was filed in January 2009. To make the assessment stick, HMRC needed to establish that the return contained a "deliberate inaccuracy" within the meaning of paragraph 4 of Schedule 24 to the Finance Act 2007 (the inaccuracy-penalty regime, which uses the same vocabulary as the discovery extension).

The case is an income-tax case, not an SDLT case. But the relevant statutory wording — "deliberate inaccuracy in a document given to HMRC by the taxpayer" — is shared across the direct-tax codes and the SDLT discovery regime, so the Supreme Court's interpretation applies directly to SDLT discovery assessments under paragraph 30(3) of Schedule 10. HMRC accepts this read-across in its published Compliance Handbook at CH53400.

What the Supreme Court actually decided

The lead judgment was given jointly by Lord Briggs and Lord Sales, with Lords Reed, Hodge and Leggatt agreeing. The Court split the question into two components:

1. Was there an inaccuracy in the document? Yes. The Court of Appeal had treated the return as a single integrated document where the white-space disclosure cured the misplaced figure. The Supreme Court rejected that reading. A return is a structured form; HMRC's processing systems read specific boxes. Putting the wrong figure in the wrong box is an inaccuracy even if the surrounding explanation is candid. (See paragraphs 51–61 of the judgment.)

2. Was the inaccuracy deliberate? No. Here the Supreme Court tightened the test. A "deliberate inaccuracy" requires that the taxpayer (or agent) intended the document to contain an inaccurate statement. It is not enough that the taxpayer deliberately filed the return; the inaccuracy itself must have been intended. (See paragraphs 47 and 62–84 of the judgment.)

Mr Tooth's accountants had explained, in the white space, exactly what they were doing. Their intention was disclosure, not concealment. The misplaced figure was a workaround forced by software limitations, not a strategy to deceive. On that finding, the assessment failed.

Critically, the Supreme Court declined to import any "blind-eye knowledge" doctrine into the deliberate-inaccuracy test. Auld's wider reading was effectively overruled on that point. To reach the 20-year window, HMRC must show actual intention — recklessness is not enough.

What HMRC has to demonstrate after Tooth

For an SDLT discovery assessment outside the four-year window to use paragraph 30(3)'s 20-year extension, HMRC must now establish, on the balance of probabilities:

1. There is an inaccuracy in the SDLT return (e.g. a stated price, a residency declaration, or a relief claim).
2. The inaccuracy caused a loss of SDLT to the Crown.
3. The taxpayer (or the agent acting for the taxpayer) intended the document to contain that inaccurate statement.

The third limb is the Tooth limb. Mistakes — even bad mistakes — do not meet it. A conveyancer who relies on the wrong slab calculation, mis-applies the additional-property surcharge, or claims first-time-buyer relief on an ineligible buyer is making an inaccuracy. Whether that inaccuracy is deliberate under Tooth depends on whether the conveyancer (or the buyer) actually intended the SDLT return to misstate the position.

In practice, that test will rarely be met by an honest taxpayer relying on professional advice. It is much more readily met where the buyer suppressed information the conveyancer would have used (for example, an undisclosed existing property triggering the surcharge) or where the price was deliberately split between consideration and "chattels" to game the slab boundary.

The six-year carelessness window in paragraph 30(2), by contrast, is a much lower bar and is the route HMRC actually relies on in most extended-window SDLT enquiries.

How Tooth fits with Hicks on "discovery"

Whether HMRC has the time to assess (the deliberate/careless analysis above) is one half of the discovery question. The other half — whether HMRC has actually discovered something — comes from Hicks v HMRC [2020] UKUT 12.

Hicks (and the line of authority running through Charlton and Sanderson) holds that a "discovery" must be a newly-appeared matter. An officer cannot re-open a settled position by re-reading the same papers and reaching a different conclusion. There has to be something new — new information, new analysis, or a newly-realised legal point — that makes the original assessment look inadequate.

Together, the two cases set the threshold for an extended-window SDLT discovery assessment:

• Hicks asks: has HMRC actually discovered something?
• Tooth asks: was the underlying inaccuracy deliberate (or merely careless)?

A challenge to a 20-year SDLT discovery assessment typically runs both arguments in parallel. For practical examples of when HMRC's later compliance-check correspondence reaches the taxpayer, see our companion piece on HMRC's CC/FS factsheet series, which is what arrives in the envelope when a discovery enquiry opens.

The penalty consequences of a "deliberate" finding

Schedule 24 FA 2007 sets the inaccuracy-penalty bands. Once an inaccuracy is classified, the penalty range tracks the behaviour:

A Tooth-style narrowing of "deliberate" therefore matters twice over: it gates HMRC's time to assess, and it caps the penalty range if an assessment is upheld. Moving a case from the "deliberate not concealed" band to the "careless" band can roughly halve the worst-case penalty.

Our explainer on Schedule 36 information-notice penalties covers the separate penalty regime that runs during an enquiry where the taxpayer fails to produce information HMRC has asked for.

What did not change

Tooth is narrow in what it actually decided. It does not change:

• The four-year ordinary window (para 30(1)) — still available without any behaviour finding.
• The six-year carelessness window (para 30(2)) — Tooth leaves the Auld line on carelessness intact. The Upper Tribunal's articulation in Auld of what amounts to carelessness by a conveyancer (broadly: a competent conveyancer would have spotted the issue) still applies.
• The "discovery" requirement itself — Hicks and Charlton survive.
• The penalty mitigation framework in Schedule 24 — unaffected; only the classification of behaviour is sharpened.

What changes is the headroom HMRC has to argue "deliberate" on facts that look more like sloppy filing than intentional misstatement.

Volume context: how many SDLT transactions sit in the higher-rate brackets

For background on the size of the SDLT return population to which these rules apply, our analysis of HM Land Registry pp-complete records (2026-05-27 query, 30.98M total transactions back to 1995, 848,775 in 2025 alone) shows that 153,233 sales in 2025 crossed the £500,000 threshold, 33,763 crossed the £925,000 5% slab boundary, and 11,725 cleared the £1.5M 12% slab — each return is potentially subject to a discovery assessment in the four-year (and on the right facts, six- or twenty-year) windows. For the underlying transaction-level data behind these figures, see our most expensive London postcodes 2025 breakdown for a worked dataset.

To explore the worked-cost stack on a high-value central London postcode, the postcode tool surfaces all current transactions, the SDLT calculation, and the all-in monthly cost: try SW1A 1AA.

Further reading on the Homecost blog

• All our Cost Intelligence pieces on SDLT, mortgage cost and the buying stack.

Sources used

Based on the Tooth v HMRC Supreme Court judgment, HMRC's SDLT Manual SDLTM56020, FA 2003 Schedule 10, FA 2007 Schedule 24, and the Hicks v HMRC Upper Tribunal decision. Land Registry transaction counts current as of the 2026-05-27 query against the pp-complete dataset.

This article is general information about how SDLT discovery assessments and the Tooth ruling work. It is not legal or tax advice. Anyone in receipt of a discovery assessment or compliance-check correspondence from HMRC should speak to a qualified solicitor, conveyancer or chartered tax adviser before responding.