This is general information about published parliamentary documents. It says nothing about whether any particular person will remain eligible for the NDIS.

What the two answers say

At the Senate Community Affairs Legislation Committee’s hearing in Perth on 6 August 2026, Senator Jordon Steele-John asked about the government’s modelling of participant exits. Two of his questions were taken on notice. The written answers, lodged in the name of the National Disability Insurance Agency and published on the committee’s inquiry page, are one paragraph each.

  • IQ26-000069, Participant Scheme Exits. Over the period to 30 June 2031, there are 108,359 assumed exits for participants with autism as a result of proposed reforms.
  • IQ26-000070, Participant Scheme Exits Below Age 9. Over the period to 30 June 2031, there are 91,707 assumed exits relating to autistic children aged 0-18 as a result of proposed reforms. Of these, 18,723 are autistic children below the age of nine.

From those three figures: 72,984 of the assumed exits relate to autistic children aged 9 to 18, and the remaining 16,652 of the 108,359 fall outside the 0 to 18 band. Children below nine are about one in five (20 per cent) of the 91,707 and about 17 per cent of the 108,359.

What the hearing added, and the caution that came with it

The Hansard transcript of the hearing, pages 56 and 57, shows where the questions came from. The figures were given by Anthea Long, a First Assistant Secretary in the Department of Health, Disability and Ageing, who said she was quoting the number that I have from the scheme actuary.

  • How the modelling worked. Ms Long said The department and the actuary did modelling based on substantially reduced functional capacity, using a severity score metric created by the NDIA, and that the modelling itself was not predicated on age or disability. Breakdowns by age and primary disability were analysis applied afterwards to those people that were seen to be potentially exited or diverted.
  • The totals. She told the committee the total number of exits to 2031 is 241,000. Of that number, 155,000 would be 18 and under. She put those 19 and over at 85,000.
  • The estimates the answers replace. At the hearing she put assumed exits around autism at around 115,000 and autistic children within the 155,000 at 95,000. The written answers are lower: 108,359, which is 6,641 fewer, and 91,707, which is 3,293 fewer. The question about children below nine was taken on notice without an estimate.
  • The caution. There is quite a lot of uncertainty around this data. And: When you start to then go down into detail into cohorts, when you start to look at particular disability types and particular ages, it becomes much, much more uncertain in terms of the confidence in that data.
  • What the modelling fed. Asked whether the figures were the basis for the projected savings, Ms Long said: Yes. The modelling informed the budget estimates.

Senator Steele-John’s question opened by citing a media report that 145,000 autistic people would lose access by 2030. The written answer gives 108,359 assumed exits for participants with autism, over a period running to 30 June 2031.

What an “exit” means here, and what it does not

Neither answer defines an exit. The hearing shows the figures are outputs of a model: Ms Long said the actuary applied a severity score to data on existing participants, and also through some sampling, and that there were projections as well that needed to be made around diversions in addition to exits. The answers give exits only, so the figures above do not include whatever the modelling projected for diversions.

Three limits follow from the documents themselves:

  • They model a threshold that was still being worked out. At the hearing Ms Long said substantially reduced functional capacity is something that we are currently consulting on, and Deputy Secretary Mary Wood said ministers had acknowledged that a technical advisory group’s advice will be the primary basis for their consideration of what to propose for the rules that will then need to be considered by the states and territories.
  • They model the bill, not the Act as passed. The answers speak of proposed reforms, and the committee set 10 August 2026 for their return. The Senate agreed 32 government amendments on 18 August and the Act received assent on 20 August, as the Parliament’s bill page records. The answers do not say whether the modelling was revisited after those amendments. What they changed is in our report on the Senate amendments.
  • They do not give timing or support needs. The answers do not say in which years the assumed exits fall, or at what point a child’s age is measured, and they say nothing about each child’s level of support needs.

The answers carry the date the committee set for their return, 10 August 2026, but no publication date.

Where Thriving Kids comes in

In the same exchange Senator Steele-John asked how many of the children would move to Thriving Kids. Ms Long did not give a number. She said Thriving Kids will be able to support children who have low to moderate autism, that The first Thriving Kids cohort is children of ages zero to eight, and that Targeted supports under Thriving Kids will be rolled out by 1 January 2028, with some supports from October 2026, to align with the access changes. Ms Wood said the access changes won’t take effect until 1 January 2028, should the states and territories agree to the work that’s been developed by the technical advisory group at the moment.

The 18,723 in IQ26-000070 is the number for that same age range, below nine. The answer does not say how many of those children have low to moderate support needs, so it cannot be read as a count of children who will move to Thriving Kids. Who the first cohort covers is set out in our reading of the National Agreement on Foundational Supports. The rollout timetable is in our Thriving Kids explainer.

Our view

This is opinion, built on the documents above. These two paragraphs are the most precise public statement we have found of what the modelling behind the bill assumes for autistic children, and they are the government’s own figures, which makes them worth having on the record. They are also exactly as uncertain as the official who gave them said: an age-and-disability breakdown of a model built on a threshold that, at the hearing, was still being consulted on.

That cuts both ways. The figures should not be quoted as a forecast of how many children will lose the NDIS. But the same modelling, on Ms Long’s evidence, informed the budget estimates, so it is the planning number. On the government’s own sequencing, Thriving Kids is to be at scale when the access changes start on 1 January 2028; the number of autistic children below nine in the modelling, 18,723, is the obvious figure to set against the number of children Thriving Kids is built to serve.

What we could and could not check

We read both answers and the hearing transcript in full at the pages cited. We have not seen the actuary’s modelling, which is not published with the answers. The NDIA’s website blocks automated reading, a constraint we have recorded since August 2026 and do not try to get around, so we make no statement about what it has published on this modelling. We have not sought comment from the NDIA or the Department.