How cases end
This is the finding worth sitting with, and it is not a criticism of anybody on its own. Of the 29,991 closed cases, 29,099 were resolved before a hearing and 892 went to a hearing and received a substantive decision.
| Outcome | Cases | Share of closed |
|---|---|---|
| Resolved by consent | 20,582 | 68.6% |
| Withdrawn | 6,019 | 20.1% |
| Dismissed | 1,912 | 6.4% |
| No jurisdiction | 424 | 1.4% |
| Decided by the Tribunal | 892 | 3.0% |
| All closed cases | 29,991 | 100% |
Extension-of-time and other categories account for the remainder and are partly suppressed in the source; the NDIA withholds counts below 11. Shares are as published.
A consent resolution means the parties agreed an outcome and the tribunal did not have to rule. Two readings are available and the data does not choose between them. It can mean the system works: a participant escalates, the agency looks again, and the matter is fixed without the cost and delay of a hearing. It can also mean decisions that would not survive scrutiny are being settled before they get any. Both are consistent with 68.6 per cent, and this table cannot separate them.
What the table does show is that adjudication is rare. Three per cent of closed cases produced a reasoned decision, which is the only part of this process that creates anything a future participant could rely on.
When the tribunal does decide
Of the 892 substantive decisions, the NDIA’s decision was affirmed in 452 and varied or set aside in 440: 134 varied and 306 set aside. On the cases that reach a ruling, in other words, the agency’s original decision stands slightly more often than not, and is changed almost half the time.
That figure needs its caveat stated rather than buried. These 892 are not a random sample of NDIA decisions. They are the cases that survived internal review, survived the pressure to settle, and were pushed all the way by someone with the capacity to do it. A near-even split on that population says nothing reliable about the other 29,099, and we are not going to pretend otherwise.
The volume, and the part that is genuinely rising
Case counts grow as the scheme grows, so the count alone proves little. The supplement publishes new cases each quarter as a share of active participants, which is the comparison that survives a growing participant base.
| Quarter ending | New cases | Share of active participants |
|---|---|---|
| 30 June 2023 | 1,064 | 0.71% |
| 30 June 2024 | 1,171 | 0.71% |
| 31 December 2024 | 1,896 | 1.11% |
| 30 June 2025 | 1,620 | 0.89% |
| 30 September 2025 | 2,483 | 1.33% |
| 31 December 2025 | 2,221 | 1.17% |
| 31 March 2026 | 2,295 | 1.20% |
| 30 June 2026 | 2,522 | 1.30% |
Selected quarters from a thirteen-quarter series running from June 2023. The June 2026 quarter is the highest on both measures.
The rate sat at 0.71 per cent in the June quarters of both 2023 and 2024, then moved up through 2024-25 and has been at or above 1.17 per cent in each of the last four quarters. At 1.30 per cent it is close to double where it was two years earlier. More participants are going to the tribunal, and not only because there are more participants.
What is being fought over
By category, planning cases are 76.0 per cent of all tribunal cases, against 19.3 per cent for access. Disputes about getting into the scheme are the minority; disputes about what is in the plan are the overwhelming majority.
| Type of support | Instances |
|---|---|
| Capacity building | 9,393 |
| Core, excluding consumables and transport | 8,029 |
| Other | 4,993 |
| General support | 2,775 |
| Assistive technology | 2,504 |
| Supported independent living | 2,435 |
| Transport | 1,598 |
| Plan management decision | 1,084 |
| Specialist disability accommodation | 887 |
| Everyday living costs | 737 |
| Home modifications | 677 |
One case can record several disputed supports, so these sum to more than the case count: 35,112 instances across 14,703 unique participants. The series begins in the March 2023 quarter, not at scheme start, so these are not lifetime totals.
Capacity building tops the list, and that is worth putting beside a second number from the same release. We reported on 17 August that capacity building is the least used support class in the scheme, at 56.6 per cent utilised against 79.2 for core. So the category participants spend least of is also the category they dispute most.
Offered as an observation, not a mechanism. These are two different tables measuring different populations over different windows, and nothing in the release connects them. A category can be both under-spent and heavily disputed for reasons that have nothing to do with each other. But capacity building is the part of a plan meant to change somebody’s trajectory rather than get them through the week, and it is simultaneously the least converted into services and the most litigated. That is a question worth putting to the agency, and we have not had it answered.
Where this sits with the rest of the release
The same release shows the NDIA completing internal reviews of its own decisions within the legislated 60 days 48.5 per cent of the time. The tribunal is the step after that one. Read together: the review pathway inside the agency misses its statutory deadline more often than not, and the pathway outside it resolves 97 per cent of closed cases without a ruling. Neither of those is evidence of a wrong decision anywhere. What they describe is a dispute system in which very little is formally determined, and a rising share of participants entering it.
Methodology
Every figure is read by us from the NDIA’s Supplement E (National) for Q4 2025-26, downloaded 17 August 2026, Tables E.82 to E.86. Percentages are as published in the file or computed from published counts and rounded to one decimal place; counts are as published.
On suppression. The NDIA withholds counts below 11 and marks some bands only as a ceiling, so several outcome categories show as fewer-than values in the quarterly columns. Where a cumulative total is published we have used that; where a figure is suppressed we have not estimated it, and the extension-of-time and other categories are therefore reported only as a remainder.
What these numbers are not. Cumulative case totals run from the start of the scheme, but the supports-in-dispute series begins in the March 2023 quarter, so the two must not be combined. The disputed-supports data is reported with a one-month lag, so the final month of the quarter is excluded. Case counts are cases, not participants: 37,027 cases involve 33,134 unique participants.
What we have not established. The release does not say how long a case takes, what a consent resolution contained, or how the 892 decided cases differ from the rest. We have not put any of this to the NDIA, and we would publish a response.
Primary sources
- NDIA, Supplement E National 2025-26 Q4 (XLSX, downloaded 17 August 2026 from the quarterly report supplements page): Table E.82 (cases by category), E.83 (cases over time incrementally and cumulatively, and as a share of active participants), E.84 (open and closed, resolved before hearing against substantive decision), E.85 (supports in dispute) and E.86 (closed cases by outcome).
- The Scheme Brief, Two NDIS deadlines are both 21 days. The agency meets one 96 per cent of the time and the other 32 (17 August 2026): the internal review figure of 48.5 per cent within 60 days.
- The Scheme Brief, NDIS plans in very remote Australia are 57 per cent spent against 75 in the cities (17 August 2026): the capacity building utilisation figure of 56.6 per cent.
Something here wrong or out of date? Tell us and we will check it against the sources above and log the outcome.