Key takeaway: SR&ED for biotech startups turns on the same two-part CRA test as any other claim: technological uncertainty, resolved through systematic investigation. But biotech runs into a line most software and hardware teams never touch, the point where experimental process development stops and commercial production starts. The CRA explicitly excludes "commercial production" and "quality control or routine testing" by name, and both show up constantly in a wet-lab workflow. Get that line right and a Canadian-controlled private corporation can still earn the enhanced 35% refundable rate on up to $6 million of qualifying spend a year, up to $2.1 million, plus lab equipment and bioreactors bought after December 15, 2024 are eligible again too.
A biotech founder asking whether their work qualifies for SR&ED usually isn't wondering if the science is real. It obviously is. Nobody spends eighteen months on an assay that doesn't work by accident. The actual confusion is narrower and harder: where does the R&D end and the production run begin? Software teams rarely have to answer that question. A biotech team answers it every time a process moves from bench to pilot scale.
This post covers what clears the bar for a company doing wet-lab experimentation, assay development, or bioprocess work: where the CRA's own exclusions land squarely on biotech workflows, what capital equipment now counts again, and where teams lose claim value by writing lab notebooks for scientists instead of for a tax reviewer.
Does SR&ED cover biotech research work
Yes. The test doesn't change because the experiment happens in a fume hood instead of an IDE. Per the CRA's eligibility guidance, work qualifies when there's a "why" (advancement of scientific or technological knowledge against genuine uncertainty) and a "how": systematic investigation by experiment or analysis. Wet-lab work fits both halves naturally. A hypothesis about why a construct isn't expressing at target yield. A controlled set of experiments testing candidate causes. A documented result, whether it worked or not. That's the exact shape of the CRA's test, and it's also just how a competent biotech lab already operates.
Biotech work sits comfortably inside the eight recognized categories of support work: engineering, design, operations research, mathematical analysis, computer programming, data collection, testing, and psychological research. Assay development is testing and design together. Bioinformatics pipelines built to make sense of your own experimental data are computer programming in service of the same underlying question. None of it needs a specific credential behind it or a specific industry label on the company.
Where the line between R&D and production actually falls
No other sector runs into this exclusion as often. The CRA's eligibility policy names "commercial production" and "quality control or routine testing" as excluded activities by name, and a scale-up program walks directly across both.
Early-stage process work is exactly the kind of technological uncertainty the program is built for: figuring out why yield drops when you move a fermentation from a 2-litre flask to a 50-litre bioreactor, or why a purification step that worked at bench scale fouls a column at pilot scale. Nobody published the answer for your specific organism, construct, and equipment train, and finding it takes structured experimentation. But once the process is characterized and repeatable, running it again to make product is commercial production, even if the batch size is still small and even if a scientist in a lab coat is the one running it. The CRA draws that line by what's actually uncertain in a given run, not by the size of the facility or the job title of the person at the bench.
Quality control follows the same split. Developing a new assay to detect a contaminant, or to measure a critical quality attribute your existing methods can't reliably catch, is experimental work: you don't know going in whether the assay will hit the sensitivity you need. Running that same assay, once validated, against every subsequent batch to confirm it meets spec is quality control or routine testing, and the CRA carves that out by name no matter how technically demanding the assay is or how much regulatory weight sits behind the release decision. The method development qualifies. Running the method afterward doesn't.
Formulation and stability work sits in the same gray zone. Screening excipients and conditions to solve an unexplained degradation pathway that no published formulation science predicts for your specific molecule is genuine investigation. Pulling samples at set intervals afterward, to confirm the finished formulation still holds against a protocol you've already validated, is not. Keep both in the same project bucket and you'll tend to overclaim the second and underdocument the first — close to the worst combination for a claim that needs to survive scrutiny.
Two biotech companies, two different claims
Consider two companies, both plausibly describing themselves as "biotech" on a pitch deck.
Company A licenses a well-characterized expression system and follows the vendor's published protocol to produce a known protein at a modest scale for a research customer. The protocol works close to as documented; the main effort goes into standing up the lab and running batches reliably. There isn't much technological uncertainty here. The biology and the process were already known before Company A started.
Company B is trying to express a difficult, novel protein construct in that same system, and it aggregates or fails to fold correctly using every published approach the team tries. They run structured experiments across expression conditions, chaperone co-expression strategies, and purification buffers, documenting what failed and why, before landing on a combination nobody had published for this construct. That documented trail of hypotheses and failed attempts is what makes Company B's claim strong, not the fact that both companies would check the same box on an industry classification form.
Capital equipment is back, and biotech feels it more than most
Yes, for capital property acquired after December 15, 2024, per the CRA's investment tax credit policy, part of the Bill C-15 changes that received royal assent in March 2026. That reversal matters enormously to a sector where a bioreactor, an HPLC system, or a specialized incubator can cost more than a year of a research associate's salary, and for years capital spend was simply excluded from SR&ED. The 35% enhanced rate applies to qualifying capital purchases too, refundable at 40%. That's narrower than the 100% refundability on salaries, but it's real money on lab equipment a biotech team was buying anyway.
In practice, that means a bench-top bioreactor bought specifically to run your scale-up experiments, or an analytical instrument acquired to develop a new assay, can enter the claim if purchased after the December 2024 cutoff and used for the experimental work itself, not for routine production runs once the process is settled. Equipment bought earlier, or equipment that's mostly running validated production batches, sits in a more limited category. Tracking acquisition dates and actual use against the $6 million expenditure limit is exactly the kind of detail that benefits from someone doing it deliberately rather than reconstructing it from purchase orders eighteen months later.
Contract research organizations and the 80% rule
Biotech teams lean on external labs more than almost any other SR&ED-claiming sector: a contract research organization running a specific assay, a specialized facility with equipment your company doesn't own, a university core lab. When that relationship is arm's length, only 80% of the contract payment enters your qualified expenditure pool, a rule that's applied since 2012 regardless of how specialized the vendor is.
Whether a CRO engagement even qualifies as a contract payment depends on the substance of the arrangement: did the CRO have to perform genuinely uncertain experimental work, or execute a well-defined protocol your team already wrote? Who bore the risk if the assay didn't work? A fixed-price validation study run to a spec you handed over looks a lot more like a service purchase than SR&ED, whatever the invoice calls it. Getting that scoping, and who owns the resulting data, settled in the contract itself matters more here than in most sectors, because biotech CRO agreements are often written by scientists focused on the science, not by anyone thinking about how a CRA reviewer will read the paper trail later.
A lot of claim value quietly disappears right here. A founder who treats the CRO invoice as a single line item ("contract research, $80,000") leaves the reviewer nothing to evaluate. Break the same engagement into what it actually paid for: a defined experimental protocol with a stated hypothesis, versus a routine analytical service run against a method that was never in question. Only the first half belongs in the claim, and separating them at invoicing time is far easier than reconstructing the split from a CRO's summary report after the fiscal year closes.
The overhead question
Salaries for the scientists and engineers running the experiments anchor most biotech SR&ED claims, and the CRA's proxy method layers an overhead allowance on top instead of forcing line-item tracking of every lab expense. The prescribed proxy amount has been 55% of your SR&ED salary base since 2014.
That 55% doesn't cleanly absorb everything a wet lab burns through. Consumables used in an experimental run, specialized reagents, and now restored capital equipment purchases each interact with the claim differently: current expenditure, capital expenditure, or already covered by the proxy. Real dollars sit on either side of that split, so it's worth resolving deliberately with whoever prepares your claim rather than assuming the proxy quietly covers it.
What biotech teams get wrong about documentation
The most common failure mode isn't ineligible science; it's real science with no evidence trail built for a tax reviewer. Lab notebooks are written to satisfy a scientist reading them next month, or a regulator years later, not to show a hypothesis, a test, and a result in the shape SR&ED needs.
A notebook entry that reads "run 14 — improved yield, moved to scale-up" tells a CRA reviewer nothing about what runs 1 through 13 tried, why they failed, or what specifically changed for run 14 to work. A reviewer isn't a biologist by default; they're reading for evidence that real uncertainty existed and that your team investigated it systematically, not for a scientifically complete methods section. Teams already generating rigorous lab records are usually sitting on most of the raw material for a strong claim. It just needs to be captured contemporaneously, separated from the pure research narrative into SR&ED's language, instead of reconstructed from a stack of notebooks after the fiscal year has already closed.
Who should think twice before claiming
Skip this, or claim narrowly, if the honest description of your work is running an established, published protocol to produce a known product, or validating an assay you already know will pass because you've run it a hundred times before. Those are real operational costs. They aren't technological uncertainty, and dressing routine production up as R&D is exactly the kind of claim that invites a CRA review it can't survive.
If your team genuinely doesn't know why a construct won't express, why a purification step falls apart at scale, or whether a new assay will hit the sensitivity a program needs, and you're testing your way through it with a documented trail, that's the program working as intended. Whether the eventual product ever reaches a regulator is beside the point.
What SR&ED for biotech startups is actually worth
The rates don't shift because the uncertainty resolved in a bioreactor instead of a codebase. A CCPC earns the enhanced 35% refundable rate on up to $6 million of qualifying expenditures a year, for tax years beginning after December 15, 2024, with the basic 15% rate applying beyond that. At the full limit, that's as much as $2.1 million annually. Current expenditures like salaries at the 35% rate are 100% refundable up to the limit for most CCPCs, cash back even against zero tax owing, while capital expenditures at that rate are 40% refundable, and excluded corporations receive a 40% refund on current expenditures instead of 100%.
The clock doesn't bend either: a corporation's SR&ED reporting deadline is 18 months after its fiscal year end, with no extension process available. A biotech team mid-scale-up, heads-down on getting a process characterized before a funding milestone, is exactly the kind of company that lets that date slip. Reconstructing why run 6 failed a year and a half ago is a lot harder than pulling a software team's training logs, because the record lives across paper notebooks, instrument printouts, and a handful of people's memory of what actually happened at the bench.
That's the part of this that doesn't care whether the R&D produced an expression construct or an API: continuous documentation, captured as the work happens, reviewed by a qualified independent SR&ED expert before anything goes to the CRA. Automation on our side handles the collecting; a named human still stands behind what's filed. It doesn't decide whether your assay or process work clears the bar — that's still the two-part test — but it means the evidence exists when the claim gets built instead of being assembled from a shelf of notebooks after the fact.
Frequently asked questions
Does biotech research automatically qualify for SR&ED? No. The CRA's eligibility test requires genuine technological uncertainty resolved through systematic investigation, the same standard applied to any software or hardware claim. Working with living systems doesn't change the test; it changes what the uncertainty and the evidence look like.
Where's the line between eligible process development and excluded commercial production? It's drawn by what's still uncertain in a given run, not by batch size or facility. Figuring out why a process doesn't scale is eligible experimental work. Running that same, now-characterized process again to make product is commercial production — explicitly excluded.
Does developing a new assay count for SR&ED, even if it's for quality control? Developing the assay can qualify if there's real uncertainty about whether it will hit the sensitivity or specificity you need. Running a validated assay against every subsequent batch to confirm spec is quality control or routine testing, which is explicitly excluded.
Can lab equipment like a bioreactor be claimed under SR&ED? Yes, for capital property acquired after December 15, 2024, used for the experimental work itself, per the CRA's investment tax credit policy. Equipment bought earlier, or used mainly for settled production runs, falls into a more limited category.
Does using a contract research organization change what I can claim? Yes, if the CRO is arm's length. Only 80% of that contract payment enters your qualified expenditure pool, and whether it counts as a contract payment at all turns on whether the CRO performed genuinely uncertain work versus executing a spec you already wrote.
Do failed experiments still count toward a SR&ED claim? Yes. The CRA's own eligibility guidance states a project doesn't need to succeed to qualify. A well-documented run that failed for a specific, understood reason is evidence of real uncertainty, not something to leave out of the claim.
Biotech doesn't get a softer SR&ED test because the work happens at a bench, and it doesn't get a harder one because a regulator might eventually look at the result. It gets the same two-part test as everything else. The only real skill is knowing exactly where your process work crosses into production.
See what your biotech research could be worth — estimate your refund or check your eligibility.