Grantwork
Insights · Eligibility

What actually qualifies for SR&ED.

Most companies that miss out on SR&ED do not miss out because their work was ineligible. They miss out because they applied the wrong test, left out the projects that failed, or swept in work the Act excludes and weakened everything around it. Here is the test the CRA actually applies, in its current form.

Published September 2025 · 9 minute read

Tworequirements, not three
35%refundable for CCPCs
$6Mannual expenditure limit
18 monthsto file, no extensions

The test is two questions, not three

Almost every article written about SR&ED eligibility says the CRA applies a three-part test. It no longer does. The policy that set out a five-question test was archived, and the guidelines that replaced it set two requirements, which the CRA itself labels the why and the how.

Work must be conducted for the advancement of scientific knowledge or for the purpose of achieving technological advancement. Work must be a systematic investigation or search that is carried out in a field of science or technology by means of experiment or analysis.The CRA's guidelines on the eligibility of work, paraphrasing the definition in subsection 248(1) of the Income Tax Act

Both have to be true of the same work. A company can be doing genuinely difficult development and still fail, if the difficulty was commercial rather than technological, or if the work proceeded by trial and error without a hypothesis behind it. The reverse is also true, and more common: a company can be doing eligible work and describe it so badly that it reads as routine.

The three questions people are thinking of are the three narrative questions on the form itself, at lines 242, 244, and 246. Those are how you report the work. They are not the test for whether it qualifies.

What technological uncertainty actually means

Uncertainty exists when you cannot know, from the knowledge publicly available in your field, whether a result can be achieved, or how to achieve it. The bar is objective. That single word decides a great many claims, because it means the question is not whether your team knew how to solve the problem, but whether the answer was available to a competent practitioner in the field.

The Tax Court has been consistent on this for decades, and it repeated the point in 2025 in a case about a concrete contractor whose appeal was dismissed. The court found no gap in the knowledge base: a person in the concrete business would be expected to know the parameters of cement creation and placement. The company's problem was real, the work was skilled, and the claim still failed, because the knowledge existed and simply had not been sought out.

Reads as uncertainty

  • A formulation that will not hold its active ingredient past a target shelf life, where the published stabilizer systems have all been tried and failed
  • A process that works at bench scale and collapses at production volume for reasons the literature does not explain
  • An architecture that hits a performance ceiling no documented pattern clears at your data volumes
  • An integration where the vendor's documented approach does not work and the reason is not discoverable

Reads as routine

  • Work that was difficult because the team was learning a technology that is well documented
  • Problems solved by hiring someone who already knew the answer
  • Configuration, tuning, and integration within known parameters
  • Development that was expensive, time-consuming, or commercially risky, but not technically unknown

Notice what is absent from the second column: failure. Failing does not disqualify a project. The CRA's own guide says the rejection of a hypothesis can be an advancement in itself, because it eliminates a possible solution. Companies routinely leave their strongest projects out of a claim because the product never shipped.

What the Act excludes outright

The definition of SR&ED ends with a list of work that is excluded no matter how difficult it was. It is worth knowing this list precisely, because most reduced claims are reduced here rather than on the uncertainty argument.

The CRA adds a plain-language set of activities that cannot be claimed: training, on-the-job learning, hiring an expert to apply what they already know, and purchasing proprietary knowledge.

Support work, and the line that trips people

Eight categories of work can be included when they support eligible research: engineering, design, operations research, mathematical analysis, computer programming, data collection, testing, and psychological research. The condition attached to them is the one to memorize, because it is where a well-intentioned claim goes wrong.

The commensurate rule
Support work is not SR&ED on its own. It can be part of the claim only when it is commensurate with the needs of, and directly in support of, the eligible research. The same activity can therefore be claimable or excluded depending on why it was done. Testing that verifies a hypothesis is support work. The same testing performed to confirm a product meets specification is quality control, and quality control is excluded.

This is why the honest scoping conversation matters more than the writing. A claim assembled by sweeping in everything the technical team touched during the year will contain excluded work, and excluded work found during a review casts doubt on the parts that were eligible.

Who can claim, and what it is worth

Any corporation carrying on business in Canada and performing eligible work in Canada can claim. A Canadian-controlled private corporation earns a 35% refundable credit on up to $6 million of eligible spending a year, which arrives as cash whether or not the company is profitable. Since the 2026 changes, eligible Canadian public corporations reach that enhanced rate too. Other corporations, including foreign-controlled ones, earn a 15% credit against tax payable.

Nearly every province adds its own credit on the same expenditures, from 3.5% to 15%, refundable in most of them. What that combination is worth in practice, with the arithmetic visible, is set out in our worked examples.

The one deadline that ends the conversation

A claim must be filed within eighteen months of the fiscal year end. The Act deems the expenditure not to be SR&ED at all if the prescribed form is not filed in time, and the CRA has no authority to extend it. There is no fairness application, no waiver, no discretion. A company that has never claimed can usually still recover its most recent completed year, and cannot recover the years before it.

Not sure which of your projects qualify?

That is the conversation. Bring a year of work and we will separate what qualifies from what does not, before anything is written.

Book a meeting Check your eligibility
Technical writing
How the narrative is actually written
CRA reviews
What makes a claim get reviewed

General information current at the date of writing, not advice on a specific claim. Program rules, forms, and administrative practice change; we confirm the current position for every file.

© Grantwork Inc.  ·  poorva@grantwork.ca  ·  grantwork.ca