Cali v National Disability Insurance Agency

This case started with a dispute about NDIS funding for early intervention therapy for a young child. Matteo Cali was four years old at the time of the Tribunal decision and had qualified for access to the NDIS because of Global Developmental Delay and Autism Spectrum Disorder Level 3. His family wanted funding for a package of early intervention supports that included Applied Behaviour Analysis therapy.

The therapy was being delivered through Autism Partnership's Little Learners Program. The practical disagreement was about how much therapy should be funded. Matteo's side sought funding for 27 hours of ABA per week. The National Disability Insurance Agency said that level was not a reasonable and necessary support under section 34 of the National Disability Insurance Scheme Act 2013 and argued that 12 hours per week was the appropriate level.

The matter had already gone through merits review in the Administrative Review Tribunal. The Tribunal did not simply ask whether the program was useful in a broad sense. It looked at the exact support sought, the evidence about outcomes, and the statutory criteria that had to be met before a support could be funded. It ultimately found that two of the required criteria were not satisfied and remitted the matter with a direction for a lower level of therapy.

The appeal to the Federal Court was therefore not a second chance to rerun the whole case on the facts. The Court made that point expressly. Its task was to decide whether the Tribunal had made an error of law.

The NDIS Act requires a decision-maker to be satisfied of several matters before a support can be funded as a reasonable and necessary support in a participant's Statement of Participant Supports. The extract sets out the relevant parts of section 34(1), including:

Section 34(1)(c), which asks whether the support represents value for money, with costs assessed relative to both the benefits achieved and the cost of alternative support.

Section 34(1)(d), which asks whether the support will be, or is likely to be, effective and beneficial for the participant, having regard to current good practice.

The Support Rules also mattered. Rule 3.1 lists matters relevant to value for money, such as whether there are comparable supports that would achieve the same outcome at a substantially lower cost, whether the support will substantially improve life stage outcomes, and whether it may reduce long-term support costs. Rules 3.2 and 3.3 deal with whether a support is effective and beneficial, including published literature, expert opinion, lived experience and what the Agency has learned through delivery of the NDIS.

The Tribunal found that some other section 34 criteria were satisfied, but not sections 34(1)(c) and (d). Because those criteria are cumulative, failure on both was enough to defeat the request for 27 hours per week. That cumulative structure became important on appeal because the applicant needed to show legal error in relation to both adverse findings.

The available reasons show that the Tribunal engaged closely with both lay and expert evidence. It considered evidence from Matteo's parents, a speech pathologist, Autism Partnership's national clinical director Ms McKinnon, and academic evidence from Dr Frazier. On the Agency side, it considered evidence from paediatrician Dr Pincus and academic Dr Sandbank.

The Tribunal also looked at the structure of the Little Learners Program and its fee schedule. The list of supports included 21 hours per week of one-to-one behaviour therapy and 6 hours per week of small-group behaviour therapy, making up the 27 hours of ABA sought. The fee schedule also included travel for home, educational setting or community visits, although the extract notes evidence that those amounts would only be drawn down if used.

Dr Pincus recommended that Matteo undertake mainstream kindergarten with supports for two days per week before moving to three days per week in the second half of 2025. The Tribunal also referred to Dr Pincus' recommendation that Matteo continue at the Little Learners Program for two days per week, together with weekly or fortnightly speech therapy, for the remainder of 2025.

Dr Frazier's evidence, as summarised by the Tribunal, focused on academic literature and supported the proposition that higher therapy hours or duration were associated with stronger outcomes for children with autism. Dr Sandbank's evidence also focused on the literature, but pointed the other way. The Tribunal highlighted her opinion that the available clinical research did not establish that 27 or more hours per week of early childhood intervention was unilaterally more effective than less intensive supports such as 5 or 15 hours per week.

The Tribunal also considered published literature, including the 2022 Australian Autism CRC guidelines. Those guidelines included recommendations that there is no set number of practitioner-delivered hours per week that leads to the best outcomes for all children, and that research evidence does not support the idea that greater amounts of support consistently lead to better child and family outcomes.

The Tribunal accepted some important parts of the applicant's case. It accepted the parents' lived experience evidence and was satisfied generally that participation in the Little Learners Program was beneficial. It also accepted that, for the parents, the program was effectively the only way to respond to Matteo's need for intensive early intervention supports, and it said there were no comparable early intervention supports that could achieve the same outcomes for him.

But the Tribunal did not stop there. It drew a distinction between the general benefit of the program and the specific question whether 27 hours per week, compared with a smaller quantum, would provide greater benefits and satisfy the statutory criteria. The Tribunal described that as the heart of the issue.

On value for money under section 34(1)(c), the Tribunal said it was generally satisfied that participation in the program was likely to reduce the cost of funding supports in the long term, but it was not satisfied on the evidence that 27 hours per week would provide greater benefits. It agreed with Dr Pincus that 12 hours a week of direct therapy represented value for money over the requested 27 hours per week.

On whether the support was effective and beneficial under section 34(1)(d), the Tribunal again accepted the parents' lived experience evidence and accepted that the program should generally be considered effective and beneficial for Matteo. However, after analysing the expert evidence and literature, it preferred Dr Sandbank's evidence that intervention intensity above 15 hours per week did not show improved outcomes compared with 5 to 15 hours per week, and it noted that Dr Pincus' views aligned with the Australian guidelines. It then concluded that it was not satisfied that 27 hours per week at the Little Learners Program would be effective and beneficial for Matteo.

The Tribunal therefore found that sections 34(1)(c) and (d) were not satisfied. It remitted the matter with a direction that Matteo's Statement of Participant Supports include 480 hours of early intervention therapy delivered by a Therapy Assistant Level 2. The Court later noted that this direction was not expressed as 480 hours at the Little Learners Program. Broken down, the 480 hours equated to 12 hours per week over 40 weeks, which was the multiplier the parties had agreed should be used.

The Federal Court was hearing an appeal under section 172 of the Administrative Review Tribunal Act 2024. The Court said clearly that its task was not to review the merits of the case. It had to decide whether the Tribunal erred in law. The extract also records that there was no contention that the Tribunal had erred in law in making any particular factual finding.

The applicant advanced two main grounds. The first concerned section 34(1)(d). He argued that the Tribunal had accepted evidence that the Little Learners Program had been effective and beneficial for him, but then later concluded that it was not satisfied that 27 hours per week would be effective and beneficial. He said that was either logically contradictory, irrational or legally unreasonable, or showed that the Tribunal had misconstrued section 34(1)(d) by effectively requiring him to prove that the requested support was more effective and beneficial than alternatives.

The second ground concerned section 34(1)(c). The applicant argued that the Tribunal had accepted there were no comparable early intervention supports that could achieve the same outcomes, but still concluded that the value for money criterion was not met. He said that was either irrational or based on a misconstruction of section 34(1)(c) by comparing the requested support with alternatives that were not actually available.

The Court also noted that because sections 34(1)(c) and (d) are cumulative requirements, the applicant needed to succeed on both grounds to succeed overall.

Button J dismissed the appeal and made no order as to costs. The available reasons show that the Court rejected the applicant's attempt to turn the Tribunal's reasoning into a legal contradiction. One of the key points made by the Court was that the applicant's argument depended heavily on reading every reference by the Tribunal to the Little Learners Program as a reference to the specific 27-hour weekly program attended by Matteo. The Court said that was not a fair reading of the Tribunal's reasons.

That point matters because it supports the distinction the Tribunal had drawn between two different propositions. The first was that participation in the program was beneficial generally. The second was whether 27 hours per week, rather than a smaller amount, met the statutory criteria for value for money and being effective and beneficial. On the available text, the Court accepted that the Tribunal had made that distinction and had not contradicted itself by doing so.

The Court also repeated the orthodox position that whether a requested support is a reasonable and necessary support is generally a question of fact for the decision-maker, subject to high thresholds for irrationality or legal unreasonableness. Mere disagreement with the Tribunal's conclusions, even strong disagreement, was not enough.

Because the available judgment text is incomplete, this page does not overstate the Court's detailed reasoning on every argument. But the orders are clear, and the extracted reasons are clear enough to show that the appeal failed because the Court did not accept that the Tribunal's approach disclosed legal error.

For most businesses, the value of this case is procedural and evidentiary. If you operate in a regulated environment, the decision shows how carefully a tribunal or regulator may separate a broad proposition from the exact proposition that matters under the statute. It may be accepted that your service, product or model is beneficial overall, but that does not automatically prove that the exact amount, scope, price or configuration you seek is legally justified.

The case also shows the importance of matching evidence to the statutory test. Here, the dispute turned on the requested dosage of therapy. Evidence that the program was beneficial in general did not necessarily answer whether 27 hours per week, compared with 12 hours per week, represented value for money or was effective and beneficial having regard to current good practice. In other regulatory settings, the same issue arises with staffing levels, service volumes, grant amounts, licence conditions, or technical specifications.

Appeal strategy is another practical lesson. Businesses often assume that if a tribunal gave too much weight to the other side's evidence, a court will fix the result. Usually that is not the court's role. If the legislation provides only an appeal on a question of law, you need to identify a legal mistake in construction, process or reasoning. If your complaint is really that the tribunal preferred the wrong expert or drew the wrong factual inference, that may not be enough.

The Federal Court judgment is dated 24 February 2026. The appeal was from an Administrative Review Tribunal decision dated 12 June 2025, notified to the parties on 27 June 2025 and amended by corrigendum on 16 July 2025. The hearing in the Federal Court took place on 10 February 2026.

The page remains in review status because the available judgment text is truncated. The result and the main reasoning points are clear enough to explain the case responsibly, but anyone publishing or relying on the case in detail should check the complete judgment first.