SR&ED Eligibility 2026: Does Your Project Qualify? CRA's Two-Requirement Test Explained

SR&ED eligibility turns on one question: did your team set out to resolve a scientific or technological uncertainty, and did it do so through a systematic investigation by experiment or analysis? If the answer is yes, part of what you spent may earn a 35% or 15% federal tax credit. If the answer is no, no amount of paperwork fixes it.

CRA no longer uses the "five questions" that many guides still repeat. Since 2021 its guidelines apply two requirements, the "Why" and the "How", taken directly from the definition in the Income Tax Act. The 2026 legislation (Bill C-15) changed the credit's limits and expenditure rules, not the definition of SR&ED, so the test is the same.

Below: the definition, both requirements with "ask yourself" checks, uncertainty vs routine engineering, support work, exclusions, software, examples, records and a self-check.

SR&ED eligibility at a glance

The legal definition of SR&ED

Everything starts with subsection 248(1). The core clause reads:

In plain terms, the work has to fall into one of three buckets:

• Basic research: advancing scientific knowledge with no specific practical application in view
• Applied research: advancing scientific knowledge with a specific practical application in view
• Experimental development: achieving technological advancement to create new, or improve existing, materials, devices, products or processes, including incremental improvements

Most business claims sit in the third bucket. Note the words "including incremental improvements": the advancement does not have to be a breakthrough. It has to be an advancement in technology, achieved through a systematic investigation, and not something your team could have done with what it already knew.

The definition then adds support work (paragraph (d)) and removes seven categories of excluded work (paragraphs (e) to (k)). Both are covered below.

CRA's two SR&ED eligibility criteria

CRA's current guidelines split the definition into two requirements. In 2021 they replaced the five questions from the 2015 policy, which CRA has archived.

The Why requirement: advancement

The work must be conducted for the advancement of scientific knowledge or for the purpose of achieving technological advancement. This is about purpose. The question is what you were trying to learn or make possible, not whether you sold more product.

• Ask yourself: what did we not know how to do at the start, in technological terms?
• Ask yourself: would a competent person in our business have known the answer, or how to get it, using standard practice?
• Ask yourself: is the goal new knowledge or capability, or is it a product feature we already knew how to build?

The How requirement: systematic investigation

The work must be a systematic investigation or search carried out in a field of science or technology by means of experiment or analysis. CRA says this includes:

• Generating an idea consistent with known facts, which serves as a starting point (a hypothesis)
• Testing the idea or hypothesis by means of experiment or analysis
• Developing logical conclusions based on the results or findings
• Keeping evidence that is generated as the work progresses

• Ask yourself: can we state the hypothesis we tested, or the approach we expected might work and why?
• Ask yourself: did we test it through experiments, prototypes, simulations or analysis, and record the results?
• Ask yourself: did the results change what we tried next, and can we show that from records made at the time?

Both requirements must be met. Systematic work aimed at a routine goal fails the Why. A real technological goal pursued by trial and error with no hypothesis or records struggles with the How.

Technological uncertainty vs routine engineering

Uncertainty is the hinge of SR&ED eligibility. CRA describes a systematic investigation as needed when it is unknown (or uncertain) whether a given result or objective can be achieved. Two details decide most cases.

First, the baseline is your business. Available knowledge is the combined scientific or technological knowledge of the resources within your business. Businesses follow known design methods, techniques, procedures, protocols, standards and other practices; work that applies those to reach a result is standard practice, not SR&ED.

Second, workarounds do not count. If you hit a problem and circumvent it with a workaround using available knowledge, CRA says that work is not eligible. The uncertainty has to be resolved, or at least investigated, not sidestepped.

CRA lists signs that available knowledge may be insufficient:

• Existing design methods are not applicable
• Requirements or specifications do not conform to existing standards
• Too many variables or unknowns
• Parameters or conditions are outside of the normal operating range
• Nature of the problem is evolving
• Data is not readily available
• There are interlocking constraints

None of these proves eligibility on its own. They are prompts. If you can point to one and explain why your team could not resolve it with known methods, you have the start of an uncertainty statement. If the accurate description is "it took a while" or "it was hard to schedule", that is difficulty, not technological uncertainty.

What a systematic investigation looks like in practice

CRA's guide to Form T661 describes an SR&ED project as a set of interrelated activities that collectively are necessary to resolve the scientific or technological uncertainties. That framing helps in two ways: it tells you what belongs in the project (everything needed to resolve the uncertainty, including qualifying support work) and what does not (activities that would have happened anyway).

A defensible investigation usually shows a clear line from a stated uncertainty, to one or more hypotheses, to tests, to results and conclusions, and then to the next round of work. The failed attempts often carry as much weight as the final answer, because they show the outcome was not known in advance.

Writing this up for Form T661 is its own skill. Our technical narrative guide covers how to turn this structure into project descriptions.

Support work: SR&ED eligible activities around the core R&D

Paragraph (d) of the definition brings in work with respect to engineering, design, operations research, mathematical analysis, computer programming, data collection, testing or psychological research, where the work is commensurate with the needs, and directly in support, of the basic research, applied research or experimental development, and is undertaken in Canada.

CRA is clear that support work is not SR&ED on its own. It rides on the back of qualifying core work. No core project, no support work.

The line between support work and excluded work is drawn by purpose. CRA's own example is data collection: if it is done to support normal business operations, it is excluded; if it is carried out to directly support the research or experimental development, it can be eligible as support work. The same test applies to testing, programming and engineering. Ask what the activity was for.

Work that never qualifies: the statutory exclusions

Paragraphs (e) to (k) of the definition exclude work with respect to:

• Market research or sales promotion
• Quality control or routine testing of materials, devices, products or processes
• Research in the social sciences or the humanities
• Prospecting, exploring or drilling for, or producing, minerals, petroleum or natural gas
• The commercial production of a new or improved material, device or product, or the commercial use of a new or improved process
• Style changes
• Routine data collection

CRA's guidelines say excluded work should only be considered after you have identified that eligible work has been performed. In other words, first find the SR&ED, then strip out what the statute removes. A common trap is commercial production: once a process works and you are running it to make saleable product, that run is production, even if the process itself came out of SR&ED.

SR&ED software eligibility

Software projects raise frequent eligibility questions. The key fact: CRA has no current software-specific eligibility guidance. Its eligibility guidance mentions software only as "computer programming", one of the support-work categories, and its page on writing T661 project descriptions includes an information and communication technology example among its samples. Software development is judged under the same Why and How test as every other field.

Applying the general test to software, the questions are the same: was there a technological uncertainty that your team could not resolve with known techniques, and did you investigate it systematically? In our view, building features with established frameworks, integrating known APIs or fixing ordinary bugs is standard practice. Investigating whether an approach can meet a performance, scale or accuracy target that existing methods do not reach may qualify, if you can show the hypothesis, the testing and the results.

On records, CRA's own list of supporting documentation includes design, system architecture and source code for software development, along with test protocols, data, results and time sheets. Version control history, commit messages and issue trackers are, in our view, useful evidence of when work happened and why, though CRA does not name them specifically.

Eligible or not? Three illustrative examples

These are simplified illustrations written by GovMoney to show how the test applies. They are not CRA examples or client cases, and a real determination depends on the facts and records.

Manufacturing

Likely eligible: a fabricator needs to weld a new alloy combination where published parameters do not apply and early trials crack. The team proposes that a modified heat input and cooling profile will prevent cracking, runs a designed set of trials, measures the results and adjusts. Likely not: the same shop sets up a new welding cell using the equipment maker's recommended settings for a standard material, then runs production and routine quality checks.

Software

Likely eligible: a SaaS company finds that known indexing approaches cannot return results within its latency target at its data volume. Engineers form hypotheses about a different data structure, build prototypes, benchmark them, record why the first two failed, and iterate. Likely not: the same company adds a reporting dashboard using its existing stack and a standard charting library, then does user testing on the layout.

Food processing

Likely eligible: a processor wants to extend shelf life without a preservative it is removing, and existing formulations do not hold texture under the new process conditions. The team tests hypotheses about ingredient interactions and processing temperatures, with lab results recorded for each trial. Likely not: the processor launches a new flavour of an existing product by adjusting seasoning to taste, redesigns the packaging, and runs routine batch testing.

Records to keep

CRA's "Get ready to claim" page says your documentation must show what work was done, who was involved, when it was done and how expenditures were calculated, and that documentation that is dated, signed and specific to the work performed is best. The guidelines add that evidence should be generated as the work progresses. Examples CRA lists:

• Planning documents, design of experiments and experimentation plans
• Project records and laboratory notebooks
• Records of trial runs
• Test protocols, data, results, analysis and conclusions
• Design, system architecture and source code (software development)
• Records of resources allocated to the project, time sheets, activity records and payroll records
• Invoices and proof of payment
• Photos, videos, prototypes, samples and scrap records

You do not submit these with the claim; the T661 asks what evidence exists. CRA may ask for it later, and claimants are responsible for the accuracy of what they file.

CRA tools that help you test eligibility

• Self-Assessment and Learning Tool (SALT): an online tool to help determine whether your work is eligible. CRA says it does not save or transfer any information to the agency.
• Pre-claim consultation: CRA's "Get ready to claim" page points businesses to a pre-claim consultation for support before filing.
• SR&ED Client Portal: a self-service space in My Business Account that includes a pre-claim workbook.
• Pre-claim approval: since April 1, 2026, eligible businesses (CCPCs, other Canadian corporations or partnerships with gross business income under $25M, in good standing) can apply for an up-front technical decision on up to three projects, with a decision within eight weeks. See our pre-claim approval guide.
• First-Time Claimant Advisory Service (FTCAS): not something you can request. CRA selects claimants (you may be selected if it is your first claim or you have not filed one in the last 3 years), and the service is only available after you file. It is mandatory if you are selected, virtual, and not a review of your claim.

SR&ED eligibility self-check

Run each project through these questions. Several "no" answers is a sign to rethink the scope or skip the project.

1. Can we name the specific technological or scientific uncertainty in one or two sentences?
2. Was the answer unknown to our team, given its combined knowledge and standard practice in our field?
3. Did we try to resolve the uncertainty, rather than work around it with known methods?
4. Can we state the hypothesis or approach we tested?
5. Did we test it by experiment or analysis, and can we show the results?
6. Did we draw conclusions and adjust the work based on them?
7. Were records created as the work happened, with dates and names?
8. Is any support work we plan to claim directly tied to this project and proportionate to it?
9. Have we removed market research, style changes, routine testing, routine data collection and commercial production?
10. Was the work, including any support work, carried out in Canada?

Common SR&ED eligibility mistakes

• Testing projects against the archived five questions instead of the current Why and How requirements
• Describing business or schedule difficulty as technological uncertainty
• Measuring uncertainty against what the claimant personally knew, rather than the knowledge available within the business
• Claiming a workaround as if it resolved the uncertainty
• Claiming support work, such as programming or testing, with no qualifying core project behind it
• Including commercial production runs once the process already works
• Relying on records reconstructed at year-end instead of evidence generated as the work progressed
• Citing software-specific CRA guidance that does not exist

Once you have confirmed which projects qualify, the next step is the claim itself. Our overview of the SR&ED tax credit covers rates, limits and eligible expenditures, and how to claim SR&ED in 2026 covers the forms and the reporting deadline.

Frequently asked questions

Does my project qualify for SR&ED?

It may if your team set out to resolve a scientific or technological uncertainty that could not be resolved with the knowledge and standard practice available in your business, and did so through a systematic investigation by experiment or analysis. Both CRA requirements, the Why and the How, must be met.

What are the SR&ED eligibility criteria in 2026?

CRA applies two requirements. The Why requirement: the work must be done to advance scientific knowledge or achieve technological advancement. The How requirement: the work must be a systematic investigation or search in a field of science or technology, by means of experiment or analysis.

Does CRA still use the five questions for SR&ED?

No. CRA replaced the five questions with two requirements in 2021, and the policy that contained the five questions is archived. Current eligibility is assessed under the Guidelines on the eligibility of work for SR&ED tax incentives.

Is software development eligible for SR&ED?

It can be. CRA has no current software-specific guidance, so software is judged under the same general test as any other field. Routine feature work and standard integrations are usually standard practice; investigating a technological uncertainty that known techniques cannot resolve may qualify.

What counts as technological uncertainty?

Uncertainty exists when it is unknown whether a given result or objective can be achieved, measured against the combined scientific or technological knowledge of the resources within your business. If you get around the problem with a workaround using available knowledge, the work is not eligible.

What work is excluded from SR&ED?

Market research or sales promotion, quality control or routine testing, social sciences or humanities research, prospecting or drilling for or producing minerals, petroleum or natural gas, commercial production or commercial use of a new process, style changes and routine data collection.

Does a failed project qualify for SR&ED?

CRA's two requirements look at the purpose of the work and how it was carried out, so missing your goal does not by itself rule the work out. Failed attempts are often strong evidence that the outcome was uncertain, provided you recorded the hypotheses, tests and results.

Can CRA tell me in advance whether my project qualifies?

Yes, through pre-claim approval, launched April 1, 2026, for eligible businesses with gross business income under $25M. CRA aims to decide within eight weeks, and approvals are valid for up to three years. The Self-Assessment and Learning Tool can also help you check eligibility on your own.