Can HMRC reopen my R&D claim after the enquiry window has closed?

Yes, but only in narrow circumstances. Once the enquiry window has passed with no enquiry opened, or an enquiry has closed, HMRC can only reopen your R&D claim through a separate power — a "discovery assessment" — and only if either the loss of tax was brought about carelessly or deliberately, or HMRC can show that a reasonably competent officer, working only from what you had actually disclosed by the relevant date, could not have realised the claim was wrong. A discovery assessment is not a general licence for HMRC to take a second, closer look at a claim it simply didn't get around to checking the first time properly.

Can HMRC Reopen a Closed R&D Claim?

The power and why it’s restricted

HMRC can raise a discovery assessment where an officer discovers that tax which ought to have been assessed hasn’t been, an assessment has become insufficient, or relief given has become excessive. Read alone, that sounds broad enough to reopen almost anything. It isn’t, because the power is deliberately hedged by two further conditions, and at least one of them has to be met before HMRC can use it against a return you’ve already filed.

The first gateway: carelessness or deliberate conduct. A discovery assessment can be made without further restriction where the loss of tax was brought about carelessly or deliberately, by you, someone acting on your behalf, or (for older partnership-adjacent structures) a partner. Where HMRC can’t establish this, it has to clear a second, higher bar instead.

The second gateway: inadequate disclosure. Where carelessness or deliberate conduct isn’t in play, HMRC can only raise a discovery assessment if a hypothetical officer — reasonably competent, with a general understanding of the law but not a specialist in your field — could not reasonably have been expected, from the information you had actually made available by the point the enquiry window closed, to realise there was a problem. This is not a test of what the actual officer who reviewed your return happened to notice. It’s a test of what a competent officer should have spotted from what was in front of them.

The practical effect is that the quality of your original disclosure is what protects a claim once the enquiry window has passed. A claim supported by a technical report that plainly and specifically explains the uncertainty being resolved, project by project, is far better protected against a later discovery assessment than one relying on vague or generic language — even where HMRC’s own review at the time was cursory.

What counts as “information made available”

This isn’t limited to the figures in the return itself. It includes the return and anything accompanying it, the claim and any documents provided with it, anything produced during an enquiry into that return, and — more broadly — anything a hypothetical officer could reasonably infer from those things, or that you specifically drew to HMRC’s attention in writing. It does not extend to material HMRC would only have found by going looking, or by piecing together information from an entirely different accounting period’s paperwork unless the connection was made reasonably obvious at the time.

Time limits

Discovery assessments run to longer time limits than an ordinary enquiry:

Four years from the end of the accounting period, as the general rule.
Six years, where the loss of tax was brought about carelessly.
Twenty years, where it was brought about deliberately, or involves a failure to comply with certain information obligations or tax avoidance notification requirements.

An objection that a discovery assessment was made out of time can only be raised on appeal against the assessment itself — it isn’t a separate, freestanding challenge.

A narrower defence: the return followed prevailing practice

There is a further, narrower protection: no discovery assessment can be made where the loss of tax is attributable to a mistake about the basis on which liability ought to have been computed, and the return was in fact made on that basis, or in accordance with the practice generally prevailing at the time. This is rarely the live issue in an R&D dispute, which is usually about the facts of a project rather than a computational basis, but it’s worth knowing it exists.

Worked example

Realbuzz Group Ltd v HMRC [2025] UKFTT 493 (TC) is the clearest illustration available of how this actually plays out for an R&D claim.

Realbuzz claimed R&D relief for its accounting period ended 30 April 2020, supported by a report covering ten projects (eight described in detail). HMRC later opened an enquiry into the following year’s return, found several of the same or related projects didn’t qualify, and — after that enquiry closed — raised a discovery assessment against the 2020 period on the view that the same problems must have existed there too. The company appealed.

The tribunal sided with the company, for reasons that turn entirely on the discovery restriction described above:

The hypothetical officer doesn’t need to know the exact number. HMRC argued the officer would have needed to work out precisely how much the claim was overstated by. The tribunal rejected this — awareness that the claim was excessive is enough; quantifying it precisely is not required.
Some projects being obviously weak is enough, even if others genuinely might have qualified. The 2020 report described several projects in terms that, on their face, described ordinary use of existing technology rather than a genuine advance — and the tribunal held that being able to spot that was sufficient to bar a discovery assessment over the whole claim, even though other projects in the same report looked more arguable.
A later year’s report doesn’t automatically count as information available for an earlier one. HMRC tried to rely on the following year’s technical report, submitted after the original return but before the enquiry window closed, as material that should have alerted a hypothetical officer to the problem in the earlier year. The tribunal disagreed — the connection between the two years wasn’t made clear enough at the time for a hypothetical officer to be expected to join the dots.
HMRC accepted there was no carelessness. This case turned entirely on the disclosure-adequacy gateway, not the carelessness gateway — the company’s adviser had genuinely engaged with the legislation and guidelines in preparing the claim.

The company kept its £335,452.57 claim.

Where claims go wrong

Treating “the enquiry closed with no changes” as final certainty. It closes the enquiry, not the file. A weak original disclosure can still be reopened within the discovery time limits if HMRC later has grounds to say a competent officer couldn’t have spotted the problem from what was actually provided.
Writing a technical report that’s vague exactly where it matters. Realbuzz shows the opposite risk too, though: where a report is specific enough that its weaker projects are honestly identifiable as weak, that specificity is what protects the stronger projects in the same claim from being reopened later.
Assuming a later year’s stronger scrutiny reopens an earlier year automatically. It doesn’t, without HMRC being able to show the connection between the two would have been clear to a hypothetical officer working from what was disclosed for the earlier year specifically.
Not distinguishing carelessness from an arguable technical position that turned out wrong. The carelessness gateway needs actual carelessness — see What penalty could apply if my R&D claim turns out to be wrong? for how that’s actually tested, which is a materially higher bar than HMRC sometimes appears to assume.
Missing the twenty-year deliberate-conduct time limit. It rarely applies, but where it does, “the claim is from years ago” is not a defence on its own.