How do I know if a client's project actually qualifies for SR&ED, and which parts of it can we claim?
A client's project qualifies for SR&ED in 2026 only if the work, done in Canada, meets both tests in the subsection 248(1) definition: it aims at a scientific or technological advancement, and it is a systematic investigation by experiment or analysis that tries to resolve an uncertainty existing knowledge couldn't. Routine testing, quality control and commercial production never qualify.
What does subsection 248(1) let your client claim?
SR&ED is a systematic investigation or search carried out in a field of science or technology by means of experiment or analysis, and it has to be basic research, applied research or experimental development laws-lois.justice.gc.ca. The CRA reads that as two requirements that must both be met: the work is done to advance scientific knowledge or achieve a technological advancement, and it's a systematic investigation by experiment or analysis. The work must also be done in Canada
canada.ca.
| Paragraph of the definition | What it covers | Claimable? |
|---|---|---|
| (a) basic research | Advancing scientific knowledge with no specific practical application in view | Yes |
| (b) applied research | Advancing scientific knowledge with a specific practical application in view | Yes |
| (c) experimental development | Technological advancement to create new or improve existing materials, devices, products or processes, including incremental improvements | Yes |
| (d) support work | Engineering, design, operations research, mathematical analysis, computer programming, data collection, testing or psychological research | Only if commensurate with the needs of, and directly in support of, (a), (b) or (c) done in Canada |
| (e) | Market research or sales promotion | No |
| (f) | Quality control or routine testing of materials, devices, products or processes | No |
| (g) | Research in the social sciences or the humanities | No |
| (h) | Prospecting, exploring or drilling for, or producing, minerals, petroleum or natural gas | No |
| (i) | Commercial production of a new or improved material, device or product, or commercial use of a new or improved process | No |
| (j) | Style changes | No |
| (k) | Routine data collection | No |
Most business claims sit in paragraph (c). The CRA's own illustration runs through all three: studying a newly discovered virus is basic research, developing a vaccine against it is applied research, and scaling up the vaccine production process is experimental development canada.ca. Getting knowledge that already exists doesn't count, so training, on the job learning, hiring experts to apply what they know and buying proprietary knowledge are all out.
What are the five questions, and do they still apply?
The CRA's 2015 Eligibility of Work policy set out a method built on five questions, answered for the work as a whole before deciding how much of it is SR&ED canada.ca. The CRA has since archived that policy and replaced it with its 2021 guidelines, which frame the test as a "why" requirement (the advancement) and a "how" requirement (the systematic investigation)
canada.ca. Both apply the same subsection 248(1) definition, so the five questions still work as a checklist for an interview with your client's technical lead.
| Question (2015 policy) | What the CRA looks for | Where claims fail |
|---|---|---|
| 1. Was there a scientific or a technological uncertainty? | Whether a result could be achieved, or how, was not known from generally available knowledge or experience | Naming the goal instead of what was missing from the knowledge base |
| 2. Did the effort involve formulating hypotheses specifically aimed at reducing or eliminating that uncertainty? | An idea consistent with known facts, set up to be proved or disproved | No hypothesis, just trying options until one worked |
| 3. Was the overall approach consistent with a systematic investigation or search, including formulating and testing the hypotheses by experiment or analysis? | Objectives and measures set early, a planned test method, logical conclusions from the results | A structured project plan mistaken for a systematic investigation |
| 4. Was the overall approach undertaken for the purpose of achieving a scientific or technological advancement? | New knowledge that could be useful beyond this project; failure can still count | Pointing to a new product or feature instead of what was learned |
| 5. Was a record of the hypotheses tested and the results kept as the work progressed? | Records made during the work, not written afterwards | Narratives rebuilt at claim time with nothing dated behind them |
The 2021 guidelines list the same building blocks for the "how" requirement: an idea consistent with known facts, testing it by experiment or analysis, logical conclusions from the results, and keeping the evidence generated as the work progresses. They add that following a structured process alone isn't enough canada.ca.
Was it a technical problem or a technological uncertainty?
This is where most reviews turn. A technical problem can be solved with the existing knowledge base, and solving it doesn't produce a technological advancement even if it produces a new product. A technological uncertainty means the solution, or the way to find it, isn't known from that knowledge base canada.ca. The CRA's example: a chemical company traces corroding pumps to a leaking upstream filter and replaces it with a known high temperature unit, which is a technical problem. The pump supplier, finding that no data exists on seal materials and profiles at those temperatures in that corrosive environment, faces a technological uncertainty.
| Fact pattern | CRA position |
|---|---|
| Cost target forces an unproven path when a costly proven one exists | The target alone creates no uncertainty, but trying to meet it can |
| Large or complex system | Size and complexity prove nothing on their own |
| Integrating known components that interact unpredictably | Can be system uncertainty, if the subsystem technologies have to change |
| Adapting a known engineering practice with reasonable certainty it will work | Development work, not SR&ED |
| Troubleshooting faults in equipment or software | Not SR&ED on its own, even when systematic; can be part of an SR&ED project |
| Novelty, a new feature or more functionality | Not an advancement unless it came from one |
| Doubt about commercial success | Not a scientific or technological uncertainty |
Each row comes from the 2015 policy canada.ca. The current guidelines say the same about cost targets and add that using a workaround based on available knowledge, instead of addressing the uncertainty, makes the work ineligible
canada.ca. Uncertainty is judged against the combined knowledge of your client's own people plus what's reasonably available publicly, so a literature search and calls to suppliers belong in the file.
Success isn't the test. Your client doesn't have to reach the goal to gain new knowledge, and learning that an idea won't work can be the advancement canada.ca. What counts is advancing science or technology, not advancing the business.
Which work around the experiments counts: support work or excluded work?
Support work under paragraph (d) isn't SR&ED on its own. It qualifies only when it's commensurate with the needs of the basic research, applied research or experimental development done in Canada, and directly in support of it canada.ca. Commensurate means matching the amount, size, extent or duration the SR&ED needed
canada.ca.
The CRA's example: a hockey stick maker integrating scanning and laser cutting works out statistically that 500 sticks will test its hypotheses at 95% confidence, then produces 2,000. Only the testing and data collection on the first 500 is commensurate and directly in support canada.ca.
Exclusions come second. The CRA first decides whether there's SR&ED at all, then uses paragraphs (e) to (k) to draw the line around it, so a project isn't disqualified because it involves, say, prospecting; only the prospecting work drops out canada.ca. Purpose separates support from excluded work: data collected to support the SR&ED can be support work, while routine data collection for normal operations is excluded even if the SR&ED team finds it useful
canada.ca.
Two more limits apply. Shop floor SR&ED done alongside production must be separated from the commercial production that paragraph (i) excludes. And the project definition doesn't let anyone split a correctly identified SR&ED project into smaller pieces that each look routine canada.ca. Define projects at the level where all the work in them is SR&ED, and make sure time and costs can be traced to each one.
Only 2,200 of 4,000 project hours are claimable in this split
Your client is the pump supplier from the CRA example above, a CCPC well under its expenditure limit. Its team logged 4,000 hours on the seal project during the year, at an average salary cost of $60 an hour, all employees who are not specified employees.
| Activity | Hours | Category | Claimable hours |
|---|---|---|---|
| Designing and running experiments on seal materials and profiles | 1,600 | (c) experimental development | 1,600 |
| Building test rigs and collecting the data the test plan called for | 600 | (d) support work, commensurate and directly in support | 600 |
| Extra test runs beyond what the test plan needed | 400 | (d) work, not commensurate | 0 |
| Quality control testing on production pumps | 500 | (f) excluded | 0 |
| Customer interviews about seal life and pricing | 300 | (e) excluded | 0 |
| Building pumps for sale with the new seal | 600 | (i) excluded | 0 |
| Total | 4,000 | 2,200 |
Claimable salary is 2,200 × $60 = $132,000. At the 35% enhanced rate, that salary alone earns $46,200 of credit, before any proxy amount. Claiming all 4,000 hours would have put $240,000 of salary in and $84,000 of credit, so $37,800 of the credit rests on hours a reviewer would remove.
The rate, the $6 million limit and the refund rules are on the SR&ED tax credit page, and where each of these numbers goes on the form is on the T661 page.
What records prove the work qualified, and can the CRA rule in advance?
The CRA wants documentation that shows what work was done, who did it, when, and how the costs were calculated. Its examples include project planning documents, design of experiments, lab notebooks, source code, records of trial runs, test protocols and results, meeting minutes, photos, prototypes, scrap records, time sheets and payroll records. Dated, signed and specific to the work is best canada.ca. The 2015 policy's fifth question is about records produced naturally during the work, not files written for the claim
canada.ca.
For a project not yet started, pre claim approval launched on April 1, 2026, with a CRA determination within eight weeks canada.ca. It's optional. The business must be a CCPC, another Canadian corporation or a partnership, have gross business income under $25 million and be in good standing, and the project can't have been in an earlier claim or involve issues under litigation
canada.ca. Work that has only just started can still apply.
The CRA's SALT tool walks through whether the work may qualify and what records a claim needs, and nothing entered in it is sent to the CRA canada.ca. Once the work qualifies, the claim still has to be filed on time: see the SR&ED filing deadline page, and the SR&ED audit page if a reviewer calls.

Sources
Income Tax Act, subsection 248(1): definition of scientific research and experimental developmentlaws-lois.justice.gc.ca
CRA: What work is eligible for SR&EDcanada.ca
CRA: Guidelines on the eligibility of work for SR&ED tax incentives (2021)canada.ca
CRA: Eligibility of Work for SR&ED Investment Tax Credits Policy, 2015 (archived, the five questions)canada.ca
CRA: Get ready to claim (supporting documentation)canada.ca
CRA: How to apply for SR&ED pre claim approvalcanada.ca
CRA: SR&ED Self Assessment and Learning Tool (SALT)canada.ca
CRA: What's new in the SR&ED program (pre claim approval launched April 1, 2026)canada.ca
Does a project that failed still qualify?
Yes, if it meets both tests. The CRA says your client doesn't have to achieve the goal to gain new knowledge, and finding that an idea won't work can be the advancement canada.ca. The 2015 policy adds that rejecting a hypothesis is advancement because it eliminates a possible solution
canada.ca.
Does software development qualify?
It can, on the same two tests as any other field. Computer programming is listed as support work in paragraph (d) laws-lois.justice.gc.ca, and the CRA lists design, system architecture and source code as supporting documentation
canada.ca. Fixing bugs or adding features with known techniques is troubleshooting or development work, not SR&ED
canada.ca.
Does a government grant for the project make it ineligible?
No. The CRA says other government funding for R&D doesn't automatically make work ineligible canada.ca. The grant reduces the expenditures your client can claim as government assistance, covered on the SR&ED tax credit page.
If some of the work was routine, is the whole project denied?
No. Exclusions only carve the routine or commercial work out of a project that already contains SR&ED, and the rest stands canada.ca. Work that looks routine in isolation can still count when it's part of a properly defined SR&ED project.
General information for the tax year shown, not advice for a specific client. Confirm the facts before you file or remit.