Back to the line
The August 2026 South Australian poll, read against the election it followed. Published . Fieldwork closed 26 August 2026, 158 days after the 21 March election. This page is updated when a figure changes and every change is listed in the update log.
Five months after the election, and in the middle of the largest protests of the term, South Australia has returned to the line it drew on 21 March. Labor is on 37 per cent, half a point from its election result; One Nation holds second at 25; the Liberals are third at 17. The June dip has closed. The measure that moved is One Nation's hold on its own voters, and the name that changed the leader question is Cory Bernardi's.
In brief
- The state is back where it voted. Labor 37, One Nation 25, Liberal 17, Greens 13, others 8: every party is within 2.6 points of its election result, and Labor is within half a point.
- The June movement has reversed. Labor rose two and One Nation fell one between June and August. Both changes are within the ±3.9 margin of error, and both run in the same direction across women, the over-55s, renters and inner Adelaide.
- The Liberals have given two points to One Nation since March and recovered none. Third place, 17 per cent, and a regional vote of 13 against One Nation's 40.
- One Nation's vote is less locked than it was. 86 per cent of its March voters are still with it, from 96 in June, and 7 per cent of them now say Labor. This is a subgroup reading and its margin is wider than the headline's.
- The park lands did not reach the poll. Fielded across the 40,000-signature petition and the 23 August rally, the survey found cost of living at 35 per cent and housing at 22 as the biggest issue, the environment at 2 and climate change at 3, and the Premier's ratings rising.
Move the preference flows yourself The lever in Act II shows how much of the 58–42 rests on where Liberal preferences go.
On every chart on this page the election result is drawn as a line, the June reading as a ring or an outline, and the August reading as a filled mark.
Where the state voted, and where it sits now
On 21 March South Australia gave Labor 37.5 per cent, One Nation 22.9, the Liberals 18.9, the Greens 10.4 and everyone else 10.3. Ten weeks later the June poll had Labor down two and a half and One Nation up three. Ten weeks after that, the August poll has Labor back on 37 and One Nation on 25: the state has moved a point or two off its result and then moved back. The line across every chart in this story is the election, and August sits on it.
Every party is within three points of its election result, and Labor is within half a point
Each row is a party: the line is its election result, the ring its June reading and the dot its August reading, with the shaded band the ±3.9 margin of error around August. The strip beneath is the two-party estimate, Labor against One Nation, against the same three readings.
| Party | Election | June | August | Since the election | Since June |
|---|---|---|---|---|---|
| Labor | 37.5 | 35 | 37 | −0.5 | +2 |
| One Nation | 22.9 | 26 | 25 | +2.1 | −1 |
| Liberal | 18.9 | 18 | 17 | −1.9 | −1 |
| Greens | 10.4 | 13 | 13 | +2.6 | 0 |
| Others | 10.3 | 8 | 8 | −2.3 | 0 |
| Reading | Labor | One Nation |
|---|---|---|
| Election, 21 March 2026 | 58.1 | 41.9 |
| June 2026 | 56 | 44 |
| August 2026 | 58 | 42 |
Labor's rise since June returns it to its result, and One Nation's fall leaves it above its result and still clear of where it started. The Liberals are the party that has not moved back, and they are third in every reading. Against the election the gains are One Nation's and the Greens' and the losses the Liberals' and the others'; against June the only movements are Labor's and One Nation's, and both sit inside the margin of error. What the chart shows is a state that moved once, in the weeks after the election, and has not moved since. The two-party figure is an estimate on both ends: the election baseline is built from a subset of seats, and the poll figure applies those same flows to the August primary votes.
ECSA result as reported by DemosAU · DemosAU, 29 May – 15 Jun 2026, n = 931 · DemosAU, 5–26 Aug 2026, n = 1,012 · Undecided excluded (7 and 5 per cent) · Two-party: DemosAU estimates from flows in seats with a Labor–One Nation final count · Changes: Ace Strategies analysis
Two parties, one estimate, and the flow that decides it
South Australia's two-party contest is Labor against One Nation, and no official state-wide count of that contest exists. DemosAU builds the figure from the preference flows it observed in the seats where a Labor–One Nation final count was held in March: Liberal voters two to one for One Nation, Greens four to one for Labor, others three to two for Labor. On those flows the election was 58.1–41.9, June was 56–44 and August is 58–42. The lever below shows how much of that figure rides on the Liberal flow.
No setting of the Liberal flow puts One Nation ahead; the headline rests on where Greens and other preferences go as much as on the Liberals'
Start from the August primary votes. The three sliders set how each party's preferences split between Labor and One Nation, with DemosAU's March estimates as the defaults: Liberals 33–67 to Labor, Greens 80–20, others 58–42. The readout states the projected two-party figure. Everything the lever shows is a projection and is labelled as one.
| Liberal flow to Labor | Labor two-party (projection) | Against the 58.1 election estimate |
|---|---|---|
| 10–90 | 53.8–46.2 | 4.3 below |
| 20–80 | 55.5–44.5 | 2.6 below |
| 33–67 · default | 57.8–42.2 | 0.3 below |
| 40–60 | 58.9–41.1 | 0.8 above |
| 50–50 | 60.6–39.4 | 2.5 above |
| 60–40 | 62.3–37.7 | 4.2 above |
Computed as Labor + Liberal × flow + Greens × 0.804 + others × 0.579 on August primaries 37 / 17 / 13 / 8; every row is a projection.
One Nation draws level only if every Liberal preference goes its way and the Greens flow falls from 80–20 to below 64–36 to Labor, with others at their default, which are flows no count has recorded. Within the range the March count supports, the Greens slider moves the figure less than the Liberal one. The published figure is computed on unrounded shares, so the lever at its defaults sits a fraction under it. The lever is a projection. It shows which assumptions the headline rests on.
Projection · August primary votes as published · Default flows: DemosAU March 2026 estimates · Ace Strategies analysis
The two points came back from women and the over-55s
The state-level movement since June is inside the margin of error. The reason to report it is that it is not evenly spread. Among women Labor rose four and One Nation fell three; among the over-55s Labor rose four and One Nation fell four; among men and the under-35s One Nation rose. Regional South Australia did not move at all: One Nation is on 40 in both waves, and it is the Liberals who fell there, from 19 to 13. Every cohort figure carries a margin wider than the headline's.
Labor gained among women, the over-55s, renters and inner Adelaide; One Nation gained among men and the under-35s
Each row is a published cohort, grouped by gender, age, education, housing and region, under South Australia as a whole. Every party is drawn on every row: the ring is June and the dot is August, so the line between them is the movement. Choose a party in the key to bring it forward and read its figures.
| Cohort | Labor | One Nation | Liberal | Greens | Others |
|---|---|---|---|---|---|
| South Australia | 35 → 37 | 26 → 25 | 18 → 17 | 13 → 13 | 8 → 8 |
| Men | 37 → 38 | 27 → 28 | 18 → 17 | 11 → 10 | 7 → 7 |
| Women | 32 → 36 | 25 → 22 | 18 → 18 | 15 → 15 | 10 → 9 |
| 18–34 | 36 → 33 | 20 → 23 | 11 → 10 | 26 → 29 | 7 → 5 |
| 35–54 | 34 → 38 | 27 → 28 | 13 → 11 | 17 → 13 | 9 → 10 |
| 55+ | 35 → 39 | 29 → 25 | 25 → 25 | 3 → 3 | 8 → 8 |
| School education | 38 → 36 | 29 → 32 | 14 → 11 | 12 → 14 | 7 → 7 |
| TAFE education small subsample | 28 → 38 | 30 → 25 | 21 → 19 | 8 → 9 | 13 → 9 |
| University education | 38 → 39 | 17 → 15 | 22 → 23 | 19 → 15 | 4 → 8 |
| Own home outright | 34 → 36 | 21 → 21 | 29 → 27 | 7 → 6 | 9 → 10 |
| Mortgage holder | 34 → 36 | 29 → 29 | 17 → 15 | 13 → 14 | 7 → 6 |
| Renter | 38 → 42 | 24 → 22 | 9 → 9 | 18 → 17 | 11 → 10 |
| Inner metro | 38 → 44 | 21 → 19 | 18 → 17 | 14 → 13 | 9 → 7 |
| Outer metro | 36 → 34 | 24 → 24 | 18 → 20 | 12 → 13 | 10 → 9 |
| Regional and rural | 25 → 28 | 40 → 40 | 19 → 13 | 12 → 11 | 4 → 8 |
The movement towards Labor is concentrated among women, the over-55s, the middle-aged, renters and inner Adelaide; the movement towards One Nation among men, the under-35s and school-educated voters. The largest single movement is among TAFE-qualified voters, on the smallest of the education subsamples; it is consistent with the rest and is not to be read on its own. Regional South Australia is the row where One Nation did not move and the Liberals fell. Every cohort figure carries a margin wider than ±3.9.
DemosAU, Jun and Aug 2026 · Subgroup margins of error are wider than ±3.9 · Published cohorts only
Outside Adelaide One Nation is three times the Liberal vote
Three small panels set One Nation against the Liberals in inner Adelaide, outer Adelaide and regional South Australia on the same scale, with June as a ring and August as a dot. The pattern follows the suburb-level map at kommshaus.dev/viz/sa-crime/ and draws no boundary finer than the three regions the report publishes.
| Region | One Nation, June | One Nation, August | Liberal, June | Liberal, August |
|---|---|---|---|---|
| Inner metro | 21 | 19 | 18 | 17 |
| Outer metro | 24 | 24 | 18 | 20 |
| Regional | 40 | 40 | 19 | 13 |
In both halves of Adelaide the two parties sit close together. In the regions One Nation is three times the Liberal vote, and the gap has widened since June because the Liberals fell. The report does not say where in the regions those points went, and this panel does not guess.
DemosAU, Jun and Aug 2026 · Three published regions · Regional subsample is the smallest of the three
The stickiest vote in the state has loosened
In June, 96 per cent of the people who voted One Nation in March said they would do so again, the highest retention of any party. In August it is 86, and the seven points that left went to Labor. Labor's own retention is 87 in both waves. The party whose voters have moved since June is One Nation, and they have moved towards the government. These are subgroup figures, each on a fifth to a third of the sample, and their margins are wider than the headline's; the movement in the One Nation row is larger than that margin.
One Nation kept 86 of every 100 March voters; in June it kept 96
Each row is a March 2026 vote and each column an August intention, so a row reads across to 100. The outlined diagonal is the share each party kept. Each cell gives the August share with the June share in smaller type, and movement the report does not itemise has its own column, which is not a party.
| March vote | Labor | One Nation | Liberal | Greens | Others | Not itemised |
|---|---|---|---|---|---|---|
| Labor | 87 → 87 | 6 → 4 | — → — | — → 3 | — → 4 | 7 → 2 |
| One Nation | — → 7 | 96 → 86 | — → 3 | — → — | — → 4 | 4 → — |
| Liberal | 5 → 5 | 5 → 7 | 89 → 85 | — → 3 | — → — | 1 → — |
| Greens | 5 → 3 | 3 → — | — → — | 89 → 94 | — → — | 3 → 3 |
| Other | 16 → 23 | 5 → 24 | 7 → — | — → 3 | 70 → 50 | 2 → — |
| Did not vote / don't remember | 18 → 32 | 21 → 22 | 12 → 4 | 27 → 23 | 22 → 19 | — → — |
Every party keeps most of its March voters, but the order has changed: in June One Nation led on retention, and in August the Greens do. The Liberal leak to One Nation has grown since June, while the One Nation leak to Labor has opened from nothing. Among people who did not vote in March or do not remember, Labor has gained and the Greens have slipped. The others row is the smallest subsample in the report and is shown for completeness rather than for its figure, and every row is a subgroup with a margin wider than the headline's.
DemosAU, Aug 2026 (p. 10) · June flows from the June report (p. 10) · Rows that do not sum to 100 carry movement the report does not itemise · Subgroup reading
The Premier, the Opposition Leader, and the name added to the question
For the first time DemosAU offered Cory Bernardi as a preferred premier alongside Peter Malinauskas and Ashton Hurn. Bernardi drew 20 per cent, Hurn 17, and Malinauskas 46, from 51 in June when the choice was two names. Undecided fell from 29 to 17. The Premier's personal rating rose three to a net +25; Bernardi's rose three to −18 and stays the lowest of the three. The party outruns its leader, as it did in June, and the leader question now measures that gap directly.
Adding Bernardi took five points from the Premier, three from Hurn and twelve from “don't know”
Each question has its own chart, because June offered three options and August four. They are not the same question, so they are drawn apart rather than as one change.
| Wave | Option | Share | Options offered |
|---|---|---|---|
| June | Malinauskas | 51 | 3 |
| June | Hurn | 20 | 3 |
| June | Don't know | 29 | 3 |
| August | Malinauskas | 46 | 4 |
| August | Bernardi | 20 | 4 |
| August | Hurn | 17 | 4 |
| August | Don't know | 17 | 4 |
The report's change column for the Premier reads +5, and the two readings in fact differ the other way; this page states both readings rather than a change, because the question changed. Malinauskas remains preferred by more than Hurn and Bernardi together. Among One Nation's own March voters the March poll had Malinauskas ahead of Hurn 32–29; the August report does not publish that cut.
DemosAU, Jun 2026 (three options) and Aug 2026 (four options) · Not a like-for-like series
All three leaders rose by one to three points; the order did not change
For each leader, the negative share runs left and the positive share right from the same line, the neutral share sits in its own column, and the net rating is printed at the end. June is the outline and August the filled bar.
| Leader | Wave | Positive | Neutral | Negative | Net |
|---|---|---|---|---|---|
| Peter Malinauskas | June | 43 | 36 | 21 | +22 |
| Peter Malinauskas | August | 44 | 37 | 19 | +25 |
| Ashton Hurn | June | 21 | 55 | 24 | −3 |
| Ashton Hurn | August | 22 | 54 | 24 | −2 |
| Cory Bernardi | June | 20 | 39 | 41 | −21 |
| Cory Bernardi | August | 20 | 42 | 38 | −18 |
Hurn's neutral share is the largest of the three: more than half the state has no view of the Opposition Leader. Bernardi's negative fell and his positive did not move, and he remains the lowest-rated of the three.
DemosAU, Jun and Aug 2026 · Net = positive − negative
What the state names as its problem, in the month it marched
The fieldwork ran from 5 to 26 August. On 19 August a petition of more than 40,000 signatures against the MotoGP and golf course works in the park lands was tabled in Parliament, and on 23 August the Stop the Chop rally was the largest of the year. The poll did not ask about the park lands. What it asked was whether the state is on the right track, and what its biggest problem is, in the respondent's own words. Two per cent said the environment and three per cent said climate change. Cost of living was 35 and housing 22, and the right-track reading rose a point.
39 right, 40 wrong, 21 unsure: the state is split as it was in June
The three answers share one scale, with June as a ring and August as a dot.
| Answer | June | August |
|---|---|---|
| Right direction | 38 | 39 |
| Wrong direction | 39 | 40 |
| Don't know | 23 | 21 |
Net sentiment is unchanged since June. In March, before the election, it was +8. The one-point moves are inside the margin of error and are reported as no change.
DemosAU, Jun and Aug 2026
Cost of living and housing are 57 per cent of the answers; the environment and climate change are five
The chart lists every August category, largest first. Cost of living and housing, the two leading categories, carry a June ring. The category set changed between waves, so no other category is compared.
| Category | August | June |
|---|---|---|
| Cost of living and inflation | 35 | 35 |
| Housing affordability and homelessness | 22 | 25 |
| Health system | 8 | — |
| Immigration and overpopulation | 7 | — |
| Crime and safety | 5 | — |
| Fuel and energy prices | 5 | — |
| State government performance | 4 | — |
| Government spending | 4 | — |
| State debt and finances | 3 | — |
| Infrastructure deficits | 3 | — |
| Climate change | 3 | — |
| Road quality | 2 | — |
| Environment | 2 | — |
| Unemployment and poverty | 2 | — |
| Education | 1 | — |
| Sustainable development | 1 | — |
| Economy | 1 | — |
| Racism and discrimination | 1 | — |
| Miscellaneous | 8 | — |
| Don't know | 2 | — |
Crime and safety sits among the second tier of answers, well behind the cost of living and housing; in Victoria DemosAU's June wave had crime at 29 per cent.
DemosAU, Aug 2026 · Single response · Comparable to June for the two leading categories only
The survey was in the field for the petition and the rally
The strip covers the 22 days from 5 to 26 August and marks three dates: the petition tabled on 19 August, the Stop the Chop rally on 23 August and the close of fieldwork on 26 August. DemosAU does not publish the daily distribution of responses.
| Date | Event | Source |
|---|---|---|
| 5 to 26 August | Fieldwork | DemosAU |
| 19 August 2026 | Petition tabled | InDaily, 19 August 2026 |
| 23 August 2026 | Stop the Chop rally | InDaily, 24 August 2026 |
| 26 August 2026 | Fieldwork closes | DemosAU |
The poll cannot say whether responses gathered after 23 August differ from those gathered before, because the report does not split them. It can say that a sample of 1,012 South Australians surveyed across those three weeks placed the environment and climate change at five per cent between them and gave the Premier a net rating three points above June's. Whether a question on the park lands themselves would have found a different picture is not something this poll can answer, and this page does not infer it.
Fieldwork: DemosAU · Events: InDaily, 19, 24 August and 2 September 2026
The strategic read
South Australia has returned to its election result. Labor's 37 and 58–42 are the March figures; One Nation's 25 is the two points it took from the Liberals in the weeks after the election and has kept; the Liberals' 17 and their regional 13 are the cost of that transfer. The movement since June is small at the state level and one-directional beneath it: women, the over-55s, renters and inner Adelaide moved towards Labor, and One Nation's own March voters loosened from 96 to 86 per cent retained. Cory Bernardi, offered for the first time, draws 20 per cent as preferred premier against a personal rating that is the lowest of the three leaders, so the party's vote and its leader's standing are as far apart as they were in June, and the question now measures the gap.
Three things will decide whether the line holds. The first is whether the seven per cent of One Nation voters now saying Labor stay there, which the next wave will show in the same flow table. The second is whether the Liberals recover any of the regional vote they have lost, from 19 to 13 in ten weeks, or whether One Nation's 40 in the regions becomes the fixed feature of the state. The third is whether an issue the poll does not ask about can move a state whose answers, in its own words, are cost of living and housing at 57 per cent and the environment at two.
Method and sources
- What this is
- This is the August 2026 DemosAU South Australian poll, commissioned in association with Ace Strategies, read against the June 2026 wave and the 21 March 2026 election result, with Ace Strategies' analysis of what moved and what did not.
- Disclosure
- Ace Strategies publishes DemosAU's polling as its research and polling partner and commissions the South Australian series. Every figure on this page is DemosAU's as published in its report; every derived figure is labelled as Ace Strategies analysis.
- The poll
- Internet panels, South Australian residents aged 18 and over eligible to vote, fieldwork 5–26 August 2026, sample 1,012, effective sample 628 after rim weighting on age by gender, education, personal income, residential tenure, language, region, 2026 state election vote and left–right self-identification. Margin of error ±3.9 points on the effective sample. Voting intention was the first substantive question after demographics. Initial undecideds were asked a leaner question; the remaining 5 per cent were excluded from the voting-intention totals.
- Two-party preferred
- No uniform state-wide two-party count was conducted at the March 2026 election. DemosAU estimates the election baseline at Labor 58.1, One Nation 41.9 by applying the preference flows observed in seats with a Labor–One Nation final count (Liberals 66.6 to One Nation, Greens 80.4 to Labor, others 57.9 to Labor) to the state-wide primary vote. The same flows applied to the August primaries give 58–42. Antony Green's Labor v non-Labor estimate of 55.7–44.3 is built from electorate-level counts and is not comparable to a state-wide poll figure. DemosAU will revise the baseline if ECSA publishes flow data; this page will follow, and the change will be logged.
- Changes since June
- Every state-level change is inside the ±3.9 margin of error and is reported as within sampling error. Direction is reported where it is consistent across cohorts. Cohort and vote-flow readings are subgroups, on smaller effective samples, and their margins are wider; the page states the caveat beside each such figure.
- The preferred premier question
- In June the options were Peter Malinauskas, Ashton Hurn and Don't know. In August Cory Bernardi was added. The two waves are not a like-for-like series for this question, and the page shows both readings without a change figure. The DemosAU report's change column reads +5 for the Premier; the June and August readings are 51 and 46.
- The biggest-issue question
- Open-ended, single response. The category set differs between June and August: June had 16 categories and August has 20, and the categories around the two leading ones changed, so only cost of living and housing are compared.
- Scenario two-party
- The preference lever applies adjustable flows to the August primary votes, with DemosAU's March estimates as defaults. It is a projection and is labelled as one wherever it appears. Computed on the rounded published primaries it gives 57.8 at the defaults; the report's 58.0 is computed on unrounded shares.
- Not included
- No question on the park lands, the MotoGP or the golf course was asked. No seat-level figure, no federal figure, no reading from any other polling house, and no daily split of the fieldwork are on this page.
- Standards
- DemosAU holds the Australian Polling Council Quality Mark and is a corporate partner of The Research Society. Its methodology statement is page 2 of the report.
- Figures
- Every figure is transcribed from the DemosAU report PDFs for August and June, which are linked from this page. Nothing is modelled except the scenario, and nothing is interpolated.
Question wording (verbatim from the report)
- Voting intention
- “If a State Election for South Australia was held today, which one of the following would you vote for?” Labor (with Peter Malinauskas as Leader); Liberal (with Ashton Hurn as Leader); One Nation (with Cory Bernardi as Leader); The Greens; Any Other Candidate; Don't Know. Answers randomly listed except Don't Know kept last. If Don't Know: “Being undecided, if a State Election for South Australia was held today, which one of the following would you have a slight leaning towards?”
- Preferred premier
- “Who is your preferred Premier?” Peter Malinauskas (Labor); Cory Bernardi (One Nation); Ashton Hurn (Liberal); Don't Know.
- Leader ratings
- “What is your opinion on the following people?” Positive, Neutral, Negative. People randomly listed, answer order randomly flipped.
- Direction
- “In your opinion, is South Australia heading in the right direction or the wrong direction?” Right Direction; Wrong Direction; Don't Know.
- Biggest issue
- “In your opinion, what is the biggest issue facing South Australia today?” Open-ended.
Questions people ask
Who is leading in South Australia?
Has the park lands protest hurt Peter Malinauskas?
What is One Nation polling in South Australia?
Who is preferred premier?
How does DemosAU calculate two-party preferred in South Australia?
When is the next South Australian poll?
Accessibility statement
The page is built to meet WCAG 2.2 at level AA. Every chart offers the same figures as a table; every instrument is reachable and operable by keyboard; every mark carries a text label; the sliders state their values in words; motion is stilled for a reader who asks their system for reduced motion. Gaps found in the audit are listed here before publication, and a fault reported to info@acestrategies.au is fixed on this page.
Update log
Every change to a figure on this page is listed here, newest first.
- The story was published on the DemosAU wave fielded 5–26 August 2026, read against the June 2026 wave and the March 2026 election result, with nine figures and one instrument. Figures were verified against both report PDFs before publication; the preferred premier readings are shown as two questions rather than one series. Komms-Haus.
The three readings and the data
| Reading | Fieldwork | n | Effective n | MoE | Undecided | Two-party | Two-party basis | Source |
|---|---|---|---|---|---|---|---|---|
| Election, 21 March 2026 | 21 March 2026 | — | — | — | — | 58.1–41.9 | DemosAU estimate | ECSA count as reported by DemosAU |
| June 2026 | 29 May – 15 June 2026 | 931 | 632 | ±3.9 | 7% | 56–44 | DemosAU 2026 flow estimates | DemosAU |
| August 2026 | 5–26 August 2026 | 1,012 | 628 | ±3.9 | 5% | 58–42 | DemosAU 2026 flow estimates | DemosAU |
The data dictionary defines every column of the CSV and every field of the JSON.
<a href="https://acestrategies.au/intelligence/back-to-the-line/">Back to the line: South Australian poll, August 2026, Ace Strategies</a>