TLDR: Broadly, you can be eligible for the R&D Tax Incentive if you are an R&D entity (a company incorporated in Australia), you conduct at least one core R&D activity meeting the four-part test in s 355-25 ITAA 1997, and you incur at least $20,000 of notional R&D deductions in the income year.
The $20,000 minimum does not apply where the expenditure is paid to a Registered Research Service Provider or is a Cooperative Research Centre contribution. Under $20 million aggregated turnover the offset is refundable at your company tax rate plus 18.5 points, commonly 43.5%.
Eligibility is self-assessed per activity against the statutory test, and registration with AusIndustry is due 10 months after year end (30 April for a 30 June year end).
The two eligibility tests
There are two separate gates, and passing one tells you nothing about the other. First, entity eligibility: is the claimant a company that the law recognises as an R&D entity, spending enough to clear the threshold? Second, activity eligibility: does the work itself meet the statutory definition of a core or supporting R&D activity?
The second test is where almost all claims are won or lost.
Entity eligibility is a checklist you can resolve in an afternoon. Activity eligibility is a judgment against worldwide knowledge, assessed activity by activity, and it is the part AusIndustry and the ATO examine when they review a claim.
Company eligibility requirements
You must be an R&D entity. In practice that means one of the following, per business.gov.au's R&D Tax Incentive guidance:
- A company incorporated under Australian law.
- A foreign company that is an Australian resident for income tax purposes.
- A foreign company resident in a country with which Australia has a double tax agreement, carrying on business through a permanent establishment here.
Sole traders, partnerships and trusts cannot claim in their own right. Nor can a company whose R&D is conducted for another entity: the claimant must bear the financial risk and hold the rights to the results.
Arrangements where trials or development run through a CRO, an associated entity, or on behalf of a foreign parent are active ATO risk areas and should be worked through carefully rather than assumed away.
The refundable offset is also unavailable to companies controlled by income-tax-exempt entities.
Minimum spend and the $20,000 threshold
You need at least $20,000 of notional R&D deductions in the income year. Notional R&D deductions are the amounts attributable to eligible activities: apportioned salary and wages plus on-costs, contractor and associate payments, directly related overheads, and decline in value of assets used for R&D.
There is one carve-out. The $20,000 floor does not apply to amounts paid to a Registered Research Service Provider, or to monetary contributions under the Cooperative Research Centre program. Those are claimable regardless of amount.
Associate payments carry an extra condition: they must be actually paid in cash by year end, not merely accrued. See what R&D expenditure you can claim for the full breakdown.
Aggregated turnover and which offset applies
Turnover does not decide whether you are eligible. It decides which offset you receive, and whether it comes as cash.
| Aggregated turnover | Offset | Rate |
|---|---|---|
| Under $20M | Refundable | Company tax rate + 18.5 pp (commonly 43.5%) |
| $20M or more | Non-refundable | Company tax rate + two-tier intensity premium |
The non-refundable premium has exactly two tiers, applied marginally: plus 8.5 percentage points on notional R&D up to 2% of total expenditure, and plus 16.5 percentage points on the portion above 2%. R&D intensity is notional R&D expenditure divided by total expenditure. There has been no flat 38.5% non-refundable rate for income years starting on or after 1 July 2021.
Worked example (refundable). A base-rate entity with $18M turnover and $600,000 of notional R&D deductions: $600,000 × 43.5% = $261,000, paid as cash to the extent it exceeds tax payable.
Worked example (non-refundable). A company with $30M turnover, $25M total expenditure and $2.5M notional R&D, assuming a 25% company tax rate. The intensity threshold is 2% of $25M = $500,000. That first $500,000 attracts 33.5% (25% + 8.5 pp), or $167,500.
The remaining $2M attracts 41.5% (25% + 16.5 pp), or $830,000. Total offset: $997,500, applied against tax payable. A company on the 30% rate would use 38.5% and 46.5% instead.
Notional R&D deductions above $150M in an income year attract the offset at the company tax rate only. The $20M threshold is legislated to rise to $50M from 1 July 2028. More detail in how to calculate your rebate.
What makes an activity a core R&D activity?
Section 355-25(1) defines core R&D activities as experimental activities whose outcome "cannot be known or determined in advance on the basis of current knowledge, information or experience, but can only be determined by applying a systematic progression of work" based on principles of established science.
That progression must proceed from hypothesis to experiment, observation and evaluation, and lead to logical conclusions. The activity must also be conducted for the purpose of generating new knowledge.
Both limbs must be satisfied. Read the words literally: paraphrasing them out of your registration is how defensibility is lost.
Outcome cannot be known in advance
The benchmark is a competent professional in the relevant field, with access to knowledge that is publicly available or reasonably accessible anywhere in the world. Not knowledge inside your company. Not knowledge inside your industry in Australia.
Pick the professional who matches the uncertainty: an ML engineer for a model architecture question, a clinician or health researcher for an efficacy question, a distributed-systems engineer for a throughput question.
Trade secrets and commercially sensitive competitor data are not reasonably accessible, so a real gap can exist even if someone has privately solved the problem. What you must show is that the answer was not in the accessible worldwide record.
That requires a prior-art search conducted before the activity started: literature, patents, internet searches, expert consultation. In Absolute Vision Technologies [2022] AATA 2319 the absence of any literature review meant no knowledge gap was demonstrated at all.
Systematic progression of work
The activity must run hypothesis, experiment, observation, evaluation, conclusions. Ad hoc iteration does not qualify.
GQHC [2024] AATA 409 held that hypotheses must be scientific, specific, and capable of being validated or invalidated. Vague commercial aims failed: a goal like building a faster platform (an illustrative example, not the tribunal's words) is not a testable hypothesis.
Active Sports Management added that the contemporaneous formation of a hypothesis is the essential starting point, and that trying things until one worked, without isolable variables, is not a repeatable experiment.
A usable hypothesis looks like this: a stated approach, a measurable threshold, a scale or population, and a disproof condition.
Purpose of generating new knowledge
The activity must be conducted for the purpose of generating new knowledge, including new or improved materials, products, devices, processes or services. Purpose is assessed on the evidence, not on intent asserted after the fact.
Commercial motive is fine. Every claimant has one. What fails is when the only knowledge produced was already in the accessible worldwide record, and is new merely to the claimant.
Supporting R&D activities
Supporting R&D activities are activities directly related to a core R&D activity, defined in s 355-30 ITAA 1997. These are the build and infrastructure work around the experiment: test harnesses, data pipelines, instrumentation, environment setup, and the engineering that makes the experiment possible.
Two conditions matter. Any supporting activity must be directly related to a core activity. And where the activity produces goods or services, or falls into one of the excluded categories, it must additionally be undertaken for the dominant purpose of supporting the core activity.
For many software companies the strongest core activity is a narrow algorithmic or systems question, with the surrounding product engineering registered as supporting. See core vs supporting R&D activities.
Activities that can never be core R&D
Section 355-25(2) lists categories that are excluded from being core activities outright. They may sometimes be supporting activities, subject to the dominant purpose test.
- Market research, market testing, market development or sales promotion, including consumer surveys.
- Prospecting, exploring or drilling for minerals or petroleum for the purpose of discovering or locating deposits, or determining their size or quality.
- Management studies or efficiency surveys.
- Research in social sciences, arts or humanities.
- Commercial, legal and administrative aspects of patenting and licensing.
- Activities associated with complying with statutory requirements or standards, including routine testing and analysis.
- Reproducing a commercial product or process from physical examination, plans, specifications or publicly available information.
- Developing or customising software for the dominant purpose of the entity's own internal administration.
From income years starting on or after 1 July 2025, activities related to gambling or tobacco cannot be core or supporting activities, unless conducted solely for harm minimisation.
Two boundaries are frequently misread. Being regulated does not trigger the compliance exclusion: work done to show compliance is excluded, experimentation that happens to occur in a regulated setting is not.
And a structured study measuring a clinical or behavioural endpoint against a pre-stated hypothesis is not market research, even when it runs on A/B infrastructure. The discriminator is the endpoint, not the tool.
Am I eligible? A self-assessment walkthrough
Work through this before you draft anything:
- Confirm the claimant is an Australian-incorporated company (or qualifying foreign resident) and bore the financial risk and holds the rights to the results.
- Identify each discrete technical or scientific hurdle you hit during the year. Do not group them: one hurdle, one candidate core activity.
- For each, name the competent professional and ask whether they could have predicted, modelled or calculated the outcome from accessible worldwide knowledge. If yes, it is not core.
- Locate the prior-art evidence that predates the activity: literature reviewed, patents searched, experts consulted, benchmarks run.
- Write the hypothesis as it existed at the time, with a measurable threshold and a disproof condition.
- Identify the observations, the evaluation against those thresholds, and the conclusions drawn.
- Check each candidate against the s 355-25(2) exclusion list and flag anything close to the line.
- Attach the supporting activities and confirm the direct relation (and dominant purpose where required).
- Total the notional R&D deductions and confirm you clear $20,000.
- Register with AusIndustry within 10 months of year end.
If a candidate activity survives steps 3, 4 and 6 with contemporaneous records, you have something defensible. If it survives only on effort and cost, you do not.
Common eligibility misconceptions
| Misconception | The actual position |
|---|---|
| "It was new to us, so it is R&D" | Novelty must be against worldwide accessible knowledge, not company experience |
| "The project failed, so we cannot claim" | Failed experiments are eligible; commercial success is irrelevant to s 355-25 |
| "We use Agile, so it is experimental" | Methodology confers nothing either way, per the AusIndustry software sector guide |
| "The whole product is the claim" | Eligibility is per activity; typically three to seven core activities |
| "It was hard and expensive" | Difficulty and cost do not create genuine technical uncertainty |
| "We are pre-revenue, so we do not qualify" | Revenue is irrelevant to eligibility; turnover only sets the offset type |
Moreton Resources Ltd v ISA [2019] FCAFC 120 confirmed that applying an existing technology in a new site or context can be a core activity where the outcome could not be known in advance. But Coal of Queensland [2021] FCAFC 54 is the counterweight: applying known methods without a genuine unknown outcome and a documented systematic progression still fails.
Eligibility by sector
The four-part test is domain-neutral. What changes is where the uncertainty lives.
Software
The eligible core is usually a specific measurable target at a stated scale that no published result demonstrates: latency, throughput, accuracy, memory, cost. Routine work (configuring documented options, integrating third-party APIs from vendor docs, building features from established patterns, refactoring, migrations) is not core. See eligibility criteria for software companies.
Health tech
The uncertainty is often clinical rather than technical. Whether an app measurably reduces IBS symptom scores is a real knowledge gap even when the software behind it is routine to build.
Adherence studies with a pre-specified behavioural or clinical endpoint can be core activities, because real-world adherence to digital interventions remains unresolved in the literature.
Trials of phases 0 to III, and pre-market pilot and pivotal stage trials, for a therapeutic good not yet on the ARTG are deemed core activities under the clinical trials determination. Phase IV (post-market) trials, trials of generic products and any market-research component are not covered. Details in clinical trial eligibility.
Hardware and embedded
Embedded control software forming part of a physical device is not caught by the internal administration exclusion. The typical core activity is whether components can be made to interoperate, or a performance envelope met, when combined behaviour is not knowable in advance.
Registration is not eligibility approval
Registering with AusIndustry does not mean your activities have been assessed as eligible. The R&D Tax Incentive is a self-assessment program. Industry Innovation and Science Australia (through AusIndustry) decides activity eligibility if and when it reviews you; the ATO administers expenditure and the offset; external merits review sits with the Administrative Review Tribunal.
The registration form reflects this. It is a field-by-field questionnaire structured as projects, then core activities, then supporting activities.
Each core activity is split across separate fields, each with its own character minimum and its own job:
- sources investigated
- why a competent professional could not know the outcome
- the hypothesis
- the experiment
- the evaluation
- the conclusions
- the new knowledge
Draft them separately.
What review actually tests
What holds a claim up under review is the evidence created while the work happened: commit histories, experiment logs, benchmark results, design decisions, protocols and ethics approvals. Absolute Vision Technologies is explicit that documents assembled after the fact, discordant with the registration, are unlikely to be accepted. Start from what evidence AusIndustry accepts.
This is the problem Rand was built for: assembling the claim from the engineering and clinical evidence your team already generates, so the activity descriptions match the record rather than a memory of it, with Expert Reviewers approving the registration before it goes to AusIndustry.
Eligibility is not a status your company holds. It is a conclusion each activity has to earn, every year, against the same four criteria. If you are unsure where your uncertainty sits, start with the complete guide to the R&D Tax Incentive and work back to the activity level from there. This article explains the rules; it is not tax advice for your circumstances.