Poll briefing · Ace Strategies

South Australia, February 2026

One Nation is on the lower-house ballot for the first time and takes 19 per cent, one point above the Liberals. Eleven of those nineteen points come out of the any-other-candidate category the October question used in its place.

State: South Australia · Ace Strategies
Fieldwork: 31 January – 16 February 2026
Sample: n=1,070 · eff. 727 · ±3.6%
Election: 21 March 2026

Labor primary

43%

Down 4 since October, and 3.0 above its 2022 result.

One Nation

19%

Its first reading on this question, against 2.6 in 2022.

Liberal

18%

Down 17.7 on 2022, and one point behind One Nation.

Eleven of One Nation's nineteen points come out of the box it was previously inside

A change between the October and February waves carries about ±5.3 points at 95 per cent, from effective samples of 661 and 727. Only the any-other-candidate movement clears it, and that category is where One Nation had been recorded.

South Australian lower-house first preferences in the October 2025 and February 2026 DemosAU polls, with the movement, in percentage points.
Party	October	February	Movement
One Nation	Not offered	19	—
Any other candidate	19	8	−11
Labor	47	43	−4
Liberal	21	18	−3
Greens	13	12	−1

The February figures exclude 6 per cent of voters who were undecided, the October figures 7 per cent. The October column comes from a question that did not list One Nation, so the movement column compares two different ballots.

The four falls sum to nineteen, which is One Nation's figure exactly. Eleven of them come from the any-other-candidate category, which is where a South Australian intending to vote One Nation had to be recorded in October. That makes this the same event New South Wales recorded in March, where thirteen of twenty-one points came from a category that fell from 20 to 7; ACE-IU-2026-025 sets out that case. Western Australia behaved differently, taking nine of seventeen points from the Liberal column instead.

So the reading is not that One Nation gained nineteen points in four months. At least eleven of them were in the October poll already, inside a category that report could not break down, and the October upper-house question had independently put the party at 12 per cent. The four points from Labor and three from the Liberals are the part that could be a movement, and neither clears its interval alone. What this wave establishes is a level, not a rise: One Nation's lower-house support in South Australia five weeks before the election was 19 per cent, with an interval of about ±2.9.

The poll declined to order One Nation and the Liberals, and the election separated them by four points

One Nation's 19 and the Liberals' 18 are one point apart, against a within-poll gap interval of about ±4.4 on the effective sample of 727. Five weeks later they finished 22.9 and 18.9.

Each bar is the 95 per cent interval on a party's February first preference, from the effective sample of 727, and the dot is the share itself.

Labor

43 39.4–46.6

One Nation

19 16.1–21.9

Liberal

18 15.2–20.8

Greens

12 9.6–14.4

Others

8 6–10

One Nation's interval runs from 16.1 to 21.9 and the Liberals' from 15.2 to 20.8. They overlap across more than four and a half points, so the poll places the two in no order, and saying which of them would form the opposition was not something this instrument could do. The election put One Nation on 22.9 and the Liberals on 18.9, a gap of 4.0 points, and One Nation became the opposition.

That sequence is worth stating carefully, because it is easy to read as a poll getting something wrong. The Liberals' final figure of 18.9 is inside this wave's interval for them. One Nation's 22.9 is one point above the top of its interval, which on a single party in a single wave is an ordinary outcome. What the poll declined to do was rank two parties whose difference it could not measure, and the election then measured it. A reader who wanted the ordering five weeks early wanted something the sample size could not supply, and ACE-IU-2026-014 shows the final pre-election wave landing every party within 1.9 points of the result and One Nation within 0.1.

One Nation takes 24 per cent of the school-educated and 8 of the university-educated

DemosAU publishes no cohort sample sizes for this wave. If the three-way splits each hold about a third of the effective sample of 727, a gap between two of them carries roughly ±7.1 points.

February 2026 South Australian lower-house first preferences by party within each demographic cohort, in per cent.
Cohort	ALP	ONP	LIB	GRN	OTH
South Australia	43	19	18	12	8
Men	44	21	19	9	7
Women	42	17	17	16	8
Aged 18 to 34	35	14	19	27	5
Aged 35 to 54	45	20	16	14	5
Aged 55 and over	45	21	19	4	11
School education	43	24	15	10	8
TAFE education	36	23	20	13	8
University education	49	8	21	15	7
Own home outright	49	16	21	5	9
Mortgage holders	41	21	20	13	5
Renters	39	19	13	20	9

Every row sums to 100 in the report. These figures are read from the demographic chart on the report's page 10, which carries no text layer. Cohort figures rest on subsamples and carry wider margins than the ±3.6% statewide topline.

Education is the division One Nation's vote follows here, as it does in Queensland and New South Wales. It takes 24 per cent of the school-educated and 8 of the university-educated, a sixteen-point gap against an interval of about ±7.1, and 23 among the TAFE-qualified. Labor runs the other way at 49 among the university-educated. The Greens' 27 per cent of under-35s against 4 among the over-55s is a 23-point gap and again the widest spread in the table.

The Liberal vote is remarkable for having no shape at all. It runs between 13 and 21 per cent across all eleven cohorts, and its best figures, 21 among the university-educated and 21 among outright owners, are three points above its statewide 18. A party whose support is that evenly thin has no cohort to defend. One Nation, by contrast, has a clear structure, and it is the structure of a party that can concentrate a vote where seats are won.

February against the 2022 election, and against the result five weeks later

Percentage points, lower-house first preferences. The last column is the 21 March 2026 result, which this wave preceded by five weeks.

South Australian lower-house first preferences at the March 2022 election, in the February 2026 DemosAU poll, and at the March 2026 election.
Party	Election, 2022	February 2026	Swing	Election, 2026
Labor	40.0	43	+3.0	37.5
One Nation	2.6	19	+16.4	22.9
Liberal	35.7	18	−17.7	18.9
Greens	9.1	12	+2.9	10.4
Others	12.6	8	−4.6	10.3

The swings are DemosAU's own, stated against its 2022 baselines, which this table reconstructs from them. The 2026 election column is the result as the June report records it. No two-party preferred figure was published for this wave; its methodology statement records the calculation method as not applicable.

The wave is closest on the Liberals, at 18 against a result of 18.9, and furthest on Labor, at 43 against 37.5. One Nation's 19 against 22.9 is 3.9 low. Those are the movements of a final five weeks of a campaign, and a wave fielded across the back half of February is not a final poll; ACE-IU-2026-014 covers the one that was.

What this wave establishes beyond doubt is the scale of what had happened to the Liberal Party. Its 17.7-point fall from 2022 is the largest fall any party records in this series, ahead of the New South Wales Coalition's 15.3, and the election confirmed it at 16.8. South Australia did not swing from Labor to the Liberals or back; it removed one of the two parties that had contested every election before it and replaced it with another, and this poll is the last full wave before the electorate said so.

Data notes

This report carries its pages as images with no text layer. The topline and the demographic table on page 10 were read from rendered pages and are the only figures cited here. The report also contains preferred premier, state direction, key person ratings, an upper-house poll, top issues, issue ratings and a deserves-to-be-re-elected question; none of those is recoverable from the file and none is quoted.

The ballot offered Labor with Peter Malinauskas as leader, the Liberals with Ashton Hurn as leader, One Nation, the Greens and any other candidate. Vincent Tarzia led the Liberals at the October wave, so the opposition changed leader between the two waves in this series.

This wave was weighted on the 2025 federal vote, where the June wave is weighted on the 2026 state vote. A change of weighting frame is one more reason to read the October-to-February movement as approximate.

Method

DemosAU South Australia State Poll, conducted in association with Ace Strategies. Fieldwork 31 January to 16 February 2026, internet panels, South Australians 18+ eligible to vote, n=1,070, effective n=727, effective margin of error ±3.6%. Rim weighting on age by gender, education, residential tenure, region, 2025 federal vote and left–right self-identification. State voting intention was the first substantive question after demographics and listed One Nation for the first time. Initial undecided voters were asked a leaner question and the remainder were excluded from the totals, which exclude 6 per cent of voters. No two-party preferred figure is published for this wave. Polling and methodology are the work of DemosAU, a member of the Australian Polling Council. This wave was commissioned in association with Ace Strategies rather than PremierNational; the analysis and presentation here remain Ace Strategies' own and the fieldwork, weighting and published figures are DemosAU's. Subgroup samples carry wider margins than the ±3.6% topline.

Use

Polling is by DemosAU, conducted in association with Ace Strategies. Material from this briefing may be reproduced in client and internal documents provided it credits acestrategies.au and the pollster. Not for resale.

Contact

Ace Strategies · info@acestrategies.au · acestrategies.au

Filing

File: ACE-IU-2026-034
Series: IU-03 · Poll briefings
State: SA · Public release
Issued: October 2026

Every poll briefing The previous South Australian wave The final poll, scored against the result The first wave of the new term