New Poll Shows Shifting Support Ahead of Israeli Elections

A new poll released by Channel 12 News, just 35 days before Israel's 26th Knesset elections, indicates significant shifts in voter preferences. If elections were held today, the "Yashr" party led by Gadi Eisenkot would emerge as the largest party with 23 mandates, maintaining its position from the previous survey. The Likud party, headed by Benjamin Netanyahu, holds steady with 20 mandates, while Naftali Bennett's "Bayit Yehudi" also retains its strength with 13 mandates.

Conversely, the "Democrats" led by Yair Golan have weakened, securing 9 mandates, while Avigdor Lieberman's "Yisrael Beiteinu" has strengthened to also receive 9 mandates. Other parties show mixed results: "Yahadut HaTorah" remains stable with 8 mandates, and the "Joint List" has weakened to 7 mandates. Shas also garners 7 mandates, "Otzma Yehudit" receives 6, and "Religious Zionism and Identity" obtains 5 mandates. Mansour Abbas's "Ra'am" party shows an increase, gaining 5 mandates. "Am Israel" receives 4 mandates, and the "Reservists and New Economy" party, led by Yoaz Hendel and Yaron Ze'ev, has again surpassed the electoral threshold, securing 4 mandates.

The poll also reveals that parties like "Blue and White" with 0.7%, "Haredi Public" with 0.3%, "Noam for Israel" with 0.1%, and "Israel First" with 0% fall below the electoral threshold. The overall bloc supporting an alternative to Netanyahu maintains 66 mandates, comprising 54 from Zionist parties and 12 from Arab parties. The current coalition holds 50 mandates, with Hendel and Ze'ev's "third bloc" party accounting for 4 mandates.

In terms of leadership suitability, Gadi Eisenkot leads Benjamin Netanyahu with 41% to 38% for Prime Minister. However, Netanyahu regains the lead when compared to Naftali Bennett, with 41% supporting Netanyahu versus 37% for Bennett. Netanyahu also leads Avigdor Lieberman, with 38% to 28%.

The survey was conducted by Midgam Consulting and Research, led by Mano Geva, with data collected on September 22, 2026. The sample included 501 respondents out of 2,498 individuals contacted, with a maximum sampling error of +/- 4.4%.