New South Wales, June 2026

01

One party moved, by eight tenths of a point more than the poll can detect

A change between the March and June waves carries about ±5.2 points at 95 per cent, from effective samples of 728 and 667. One Nation's rise of 6 is the only movement that clears it, and it clears it narrowly.

Each bar is the change in a party's first preference from March to June 2026, in points. The pale band is the ±5.2 points a change of that size carries.

A six-point move against an interval of ±5.2 is a finding, but it is a finding with very little room in it. One further wave in either direction would settle it; the August wave, which puts One Nation back on 23, does not, because that four-point fall is itself inside the interval. What the two waves together support is that One Nation's New South Wales vote sits somewhere in the middle twenties, and that it has been there since March.

The order at the top of the ballot is a separate question, and the poll does not settle it either. Labor's 32 and One Nation's 27 are five points apart, and a within-poll gap of that size on an effective sample of 667 carries about ±5.8. One Nation's seven points over the Coalition do clear their interval of about ±5.2, so the statement this wave supports is that the Coalition is third, not that Labor is first.

02

One Nation leads Labor by fourteen points in regional New South Wales

DemosAU states its cohort subsamples as approximately 200 to 500, which makes these intervals computed rather than assumed. A gap between two cohorts carries roughly ±6.5 to ±8.6 depending on their size, and a gap between two parties inside one cohort roughly ±8 to ±10.6.

Every row sums to 100 in the report. The Coalition's share sits between 19 and 21 in all five cohorts, so none of the variation here is its.

One Nation's regional 37 against its inner-metropolitan 15 is a 22-point gap that clears its interval on any plausible pair of subsample sizes. Inside the regional cohort its fourteen points over Labor clear the roughly ±8 to ±10.6 a within-cohort gap carries, so One Nation leading Labor in regional New South Wales is a finding of this poll rather than an artefact of a small cell. Outer metropolitan is the boundary: Labor and One Nation are level on 29 there, and the poll places them in no order.

Language divides the sample as sharply as geography. Among voters who speak only English at home One Nation takes 31 per cent and leads Labor; among those who speak another language it takes 12 and Labor takes 43. The nineteen-point difference in One Nation's share clears an interval of about ±7.3, and the two cohorts are not independent of the geographic ones: the inner-metropolitan and other-language rows record the same 43 for Labor and the same 20 for the Coalition, and sit three points apart on One Nation. What the two cuts describe together is a single division rather than two.

The Coalition is the party this section cannot find a base for. Its vote runs from 19 to 21 across every location and language cohort in the table, which at 20 per cent statewide means it is uniformly weak rather than regionally concentrated. A party polling its statewide figure in every segment of the electorate has no cohort to defend and none to attack from.

03

The government's rating improved on all seven issues while its vote did not

Asked whether the Minns government is doing a satisfactory job on each issue, respondents rated it better in June than in February on every one, by between 2 and 11 points. Its primary vote over the same period fell 2.

Each bar is the net rating on an issue, the satisfactory share less the unsatisfactory share, in points. The note gives the three underlying shares and the movement since February.

Every row improved and the government lost ground anyway. The two issues voters name most often as the biggest facing the state, cost of living at 41 per cent and housing at 23, are the two the government is rated worst on, at 55 and 47 points negative. An eleven-point improvement on energy, from a net of 32 negative to 21 negative, is a real movement on a question nobody in this poll names as their first concern. Improvement where it does not count and a deficit where it does is a coherent account of a government whose satisfaction ratings and vote share are moving in opposite directions.

The don't-know shares are worth reading alongside the nets. On managing the state budget 38 per cent of respondents have no view, more than rate it either way, so the net of 12 negative rests on the 62 per cent who answered. On cost of living only 19 per cent have no view. Voters are decided about the things that reach them and undecided about the things that do not, which is the ordinary shape of this question and a reason to weight the first two rows more heavily than the rest.

04

One Nation runs eighteen points higher among the TAFE-qualified than the university-educated

On DemosAU's stated subsample range of approximately 200 to 500, a gap between two cohorts carries roughly ±6.5 to ±8.6 points.

Every row sums to 100 in the report. DemosAU advises caution on cohort figures and on changes between waves within a cohort, given subsamples of approximately 200 to 500.

Education is the sharpest division. One Nation takes 33 per cent of the TAFE-qualified and 15 of the university-educated, an eighteen-point gap, while Labor runs 40 among the university-educated and 27 among the school-educated. Age runs the same way for One Nation, from 15 per cent of under-35s to 33 of the over-55s, and the Greens run against it, from 25 to 6. Each of those gaps clears the interval its cohorts support.

One Nation's eight-point lead among women over men, 31 against 23, sits just outside an interval of about ±6.7. It is the second wave in a row to show the gap in that direction, after March's seven points, and it is the one finding in this series that most warrants a third wave before anyone builds on it. The Coalition's best cohort in the table is renters at 26 per cent, which is six points above its statewide figure and its only reading above 25 anywhere in this wave.

05

The upper house projects six One Nation members, and no two-party figure is published

One Nation takes 27 per cent of the upper-house vote, against 30 for Labor and 20 for the Coalition, and DemosAU projects it into six of the twenty-one seats contested at the next election.

The projection is DemosAU's and carries the uncertainty of the shares behind it. The final column is the chamber as DemosAU records it now, where Mark Latham sits among the four independents rather than in a row of his own. Both projection columns sum to 21 and both chamber columns to 42.

In October 2025 the same exercise gave One Nation three seats and left five in doubt. Eight months later it gives the party six and leaves two in doubt, and the doubt has resolved towards One Nation rather than away from it. A chamber of 14 Labor, 11 Coalition and 6 One Nation is one in which the government needs eight votes it does not have and the opposition needs eleven, and in which One Nation holds the largest single bloc outside the majors.

This wave publishes no two-party preferred figure. Its methodology statement records the calculation method as not applicable, where the March wave published 61–39 with a written caution. That is a defensible retreat rather than an omission: with One Nation on 27 per cent and 62 per cent of its preferences exhausting under optional preferential voting in March's flows, a two-party number would be struck on about three-quarters of the ballot and would move with the flow assumption more than with opinion. ACE-IU-2026-025 sets out that arithmetic.