On Census night in August 2021, 122,494 people in Australia were counted as homeless, 48.2 for every 10,000 people. In 2024–25, 288,970 people received help from specialist homelessness services. The two figures are not estimates of the same thing, and neither is a count of people sleeping out: sleeping out, in tents or in improvised dwellings accounts for 6.2% of the Census count, while the largest group, 39.1%, is people living in dwellings that need four or more extra bedrooms. The Census count has risen at every Census since 2006, but part of the 2021 count is not directly comparable with earlier ones: excluding the two groups the ABS flags, the count fell between 2016 and 2021, and the rate per head fell too.
The Census rate is per 10,000 people counted on one night; the service rate is per 10,000 estimated resident population over a year, so the two are not comparable. Node reading for Housing Policy (Affordable, Social) (21330): NIL RETURN. This article rests on the source data alone.
What is counted as homelessness
The Australian Bureau of Statistics (ABS) defines homelessness by the absence of accommodation alternatives, not by the absence of a roof. A person is homeless if they have no suitable alternative and their current living arrangement is in a dwelling that is inadequate, has no tenure or a short tenure that cannot be extended, or does not allow them control of, and access to, space for social relations. It estimates the number on Census night by placing people into six mutually exclusive groups, so that each person appears once.
| Group | Who is included | 2021 | Share |
|---|---|---|---|
| Improvised dwellings, tents or sleeping out | People with no dwelling, in tents, or sleeping out | 7,636 | 6.2% |
| Supported accommodation for the homeless | People in accommodation provided by specialist homelessness services | 24,291 | 19.8% |
| Staying temporarily with other households | People with no usual address staying with another household | 16,597 | 13.5% |
| Boarding houses | People usually resident in a boarding or rooming house | 22,137 | 18.1% |
| Other temporary lodgings | People with no usual address in hotels, motels and other temporary lodging | 3,934 | 3.2% |
| Crowded dwellings needing four or more extra bedrooms | People in dwellings needing four or more extra bedrooms to house their occupants adequately | 47,895 | 39.1% |
| All homeless persons | 122,494 | 100.0% |
People counted on Census night, 10 August 2021, by place of usual residence. Groups do not sum exactly to the total because the ABS randomly adjusts cells to protect confidentiality. Source: ABS, Estimating Homelessness: Census, 2021, tables 1.1 and 1.3.
The ABS separately counts people in other marginal housing. The largest group, 81,026 people in 2021, were living in dwellings needing three extra bedrooms, one short of the threshold. They are close to the definition but outside it, and are not in the homeless count.
Where the Census count sits
Sleeping out, in tents or in improvised dwellings is one of the smaller groups: 7,636 people, 6.2% of the total. The crowding group has been the largest at every Census since 2006 and accounted for 47,895 people in 2021. Supported accommodation, boarding houses and staying temporarily with other households each accounted for between 13% and 20%.
People counted as homeless on Census night, by group, 2021
Per cent of the 122,494 people counted; groups as defined in the table above. Source: ABS, Estimating Homelessness: Census, 2021, table 1.1.
Four Censuses
The count has risen at every Census: 89,733 in 2006, 102,439 in 2011, 116,427 in 2016 and 122,494 in 2021. Measured against the population, the rate rose from 45.2 to 49.8 per 10,000 and then fell to 48.2 in 2021.
The 2021 figures for two groups are not directly comparable with earlier Censuses. The ABS states that this reflects “improvements in data quality through greater use of administrative data”. The boarding house group rose from 17,503 to 22,137, and other temporary lodgings from 678 to 3,934, partly for that reason. The four groups the ABS does not flag together fell from 98,248 in 2016 to 96,419 in 2021; the crowding group fell by 3,193, more than the net fall across the four.
| Group | 2006 | 2011 | 2016 | 2021 |
|---|---|---|---|---|
| Improvised dwellings, tents or sleeping out | 7,252 | 6,810 | 8,200 | 7,636 |
| Supported accommodation for the homeless | 17,334 | 21,258 | 21,235 | 24,291 |
| Staying temporarily with other households | 17,661 | 17,374 | 17,725 | 16,597 |
| Boarding houses (a) | 15,459 | 14,944 | 17,503 | 22,137 |
| Other temporary lodgings (a) | 501 | 682 | 678 | 3,934 |
| Crowded dwellings needing four or more extra bedrooms | 31,527 | 41,370 | 51,088 | 47,895 |
| All homeless persons | 89,733 | 102,439 | 116,427 | 122,494 |
People counted on Census night. (a) 2021 not directly comparable with earlier Censuses, owing to greater use of administrative data. Source: ABS, Estimating Homelessness: Census, 2021, table 1.1.
By state and territory
Among the states and the Australian Capital Territory, the 2021 rate ranged from 36.6 per 10,000 in Western Australia to 47.1 in Victoria. Victoria’s rate includes the change in the boarding house count: 8,599 people in Victorian boarding houses in 2021, against 4,406 in 2016, most of the national rise in that group. The mix of circumstances differs more than the rates do. New South Wales and Victoria together account for 79% of people counted in boarding houses. In Western Australia, sleeping out or in improvised dwellings is 23.8% of the count, the highest share of any state or territory.
The Northern Territory’s rate, 563.6 per 10,000, is of a different order, and it is driven mainly by one group: 9,904 of the 13,104 people counted there, 75.6%, were living in dwellings needing four or more extra bedrooms. It is shown in the table and omitted from the chart, where it would compress every other bar.
People counted as homeless per 10,000 population, 2021
Rate per 10,000 people, by state or territory of usual residence. The Northern Territory (563.6) is omitted for scale. Source: ABS, Estimating Homelessness: Census, 2021, table 1.5.
| State or territory | Homeless persons | Rate per 10,000 | In crowded dwellings (four or more extra bedrooms) |
|---|---|---|---|
| New South Wales | 35,011 | 43.4 | 14,640 |
| Victoria | 30,660 | 47.1 | 9,068 |
| Queensland | 22,428 | 43.5 | 7,839 |
| South Australia | 7,428 | 41.7 | 2,352 |
| Western Australia | 9,729 | 36.6 | 2,935 |
| Tasmania | 2,350 | 42.1 | 709 |
| Australian Capital Territory | 1,777 | 39.1 | 447 |
| Northern Territory | 13,104 | 563.6 | 9,904 |
| Australia | 122,494 | 48.2 | 47,895 |
Census night, 10 August 2021, by state or territory of usual residence. Source: ABS, Estimating Homelessness: Census, 2021, tables 1.3 and 1.5.
Specialist homelessness services
The Australian Institute of Health and Welfare (AIHW) counts something different: people who received support in a year from agencies funded to provide specialist homelessness services. Its count includes people who were homeless when support began and people at risk of homelessness, and excludes anyone who did not seek or did not receive support. It is a measure of service use, and it depends on how many agencies there are and what they can provide.
Clients numbered 236,429 in 2011–12, when the collection began, and 288,970 in 2024–25; the highest count was 290,462, in 2019–20. Per head of population, use was 105.8 per 10,000 in the first year, peaked at 119.2 in 2016–17 and was 106.3 in 2024–25.
Of clients whose situation was recorded, 48.7% were homeless at first presentation in 2024–25, and the rest at risk. The number homeless at first presentation has risen in each of the last three years, from 113,815 in 2021–22 to 132,542 in 2024–25, while the number at risk moved from 134,468 to 139,489. Among those homeless at first presentation, 34,833 had no shelter or were in an improvised or inadequate dwelling, and 55,028 were in short-term temporary accommodation.
Clients of specialist homelessness services per 10,000 population, 2011–12 to 2024–25
Clients per 10,000 estimated resident population; each point is a financial year, labelled by the year in which it begins. Source: AIHW, Specialist homelessness services annual report 2024–25, historical table HIST.CLIENTS.
Why the two counts differ, and what neither shows
The Census counts people homeless on one night under a statistical definition. The service count covers a whole year, includes people at risk, and misses people who do not approach a service. The two overlap, because the Census group for supported accommodation includes people housed by specialist homelessness services on Census night, but neither is contained in the other, and the difference between them is not itself a measure. The next Census-based estimates will come from the 2026 Census; the ABS has not yet published a release date.
Neither headline count measures how long people remain homeless, or whether the same people are counted from one Census to the next. The Census relies on people being counted on a single night. The service count can rise and fall with service capacity as well as with need. How social housing waiting lists count applicants is set out in What a Social Housing Waiting List Counts.
Findings are presented on the basis of data and evidence alone.
