Will AfD win a plurality of seats in the next German federal election?

TL;DR

I assign a 74% chance that AfD wins the strict individual-party seat plurality. The latest poll from each of nine institutes averages 27.6% for AfD and 20.8% for CDU/CSU; separating CSU leaves CDU near 16–17%, roughly 11 points behind AfD (DAWUM; Federal Returning Officer). The lead is large, but the ordinary election is not due until early 2029, leaving time for a CDU leadership reset, voter consolidation or an AfD-specific shock (Bundestag).

Context

The official final result of the February 23, 2025 election gave CDU 164 seats, AfD 152, SPD 120, Greens 85, the Left 64, CSU 44 and SSW one. AfD was therefore only 12 seats and 1.75 percentage points of second votes behind CDU as an individual party, despite trailing CDU and CSU combined by much more (Federal Returning Officer).

The current system fixes the Bundestag at 630 seats and makes qualifying parties’ second votes the main determinant of party seat totals; constituency victories do not create extra party seats without second-vote coverage (Bundestag). The present Bundestag first met on March 25, 2025, so the regular election falls in early 2029 unless the Bundestag is dissolved sooner (Bundestag).

Evidence

AfD has contested four federal elections. Its full federal history shows a steady closing of the gap with CDU, though four observations are too few to extrapolate mechanically.

The main warning comes from the forecast horizon. The usable polling archive supplies only two clean analogues about 900 days before an election. In spring 2019, CDU/CSU polled near 29–31%, SPD around 15–18%, Greens around 17–21% and AfD around 12–14%; the 2021 election instead produced SPD 25.71%, CDU 18.95%, Greens 14.72% and AfD 10.39% (DAWUM archive; 2021 result). In September 2022, polls generally put CDU/CSU around 24–30%, Greens around 19–24%, SPD around 17–22% and AfD around 11–14%; by 2025 AfD had risen to 20.80% while the Greens had fallen to 11.61% (DAWUM archive; 2025 result). Large reversals over this distance are normal enough to matter.

The present polling signal is still unusually strong. Taking the latest poll from each of nine institutes published between August 18 and September 5, 2026 gives an institute-balanced average of AfD 27.6% and CDU/CSU 20.8%. Every one of those polls places AfD ahead of the combined Union, with margins ranging from five to eight points and samples ranging from 1,039 to 2,504 respondents (DAWUM). The September ARD DeutschlandTrend independently recorded AfD at 27% and the Union at 21%, using 1,319 interviews conducted from August 31 through September 2 (Infratest dimap).

From 2002 through 2025, CDU averaged about 79.6% of the combined CDU/CSU second vote. Applying that split to 20.8% gives CDU roughly 16.6%, leaving AfD about 11 points ahead (Federal Returning Officer historical results). This may understate the current gap: an August 2026 Bavarian state poll had CSU at 38%, close to its 37.18% federal result in Bavaria in 2025, which would imply that CSU retains nearer six national percentage points and CDU is closer to 15% (DAWUM Bavaria; 2025 Bavaria result). I retain the less favorable historical proportional split in the model.

Actual elections support AfD’s rise but also show its regional limits. AfD won a preliminary 43.79% and 39 of 83 seats in Saxony-Anhalt on September 6, 2026, while CDU received 17.2%; turnout was 77.82% (Saxony-Anhalt election authority). Yet AfD received 18.8% in Baden-Württemberg and 19.5% in Rhineland-Palatinate earlier in 2026, finishing well behind CDU or the Greens in both western states (Baden-Württemberg; Rhineland-Palatinate). The Saxony-Anhalt result proves that high AfD polling can translate into votes. It does not support a 44% national forecast.

AfD combines strong retention with a limited expansion pool. Ipsos reports that 96% of its 2025 voters would currently vote for it again, while the Union retains only slightly more than half of its former voters and loses more than one in seven to AfD (Ipsos). Conversely, a June Forsa survey found that only 8% of current non-AfD voters could imagine supporting it, while the Union’s stated potential electorate was 45% (RTL/Forsa). I read this as a sticky AfD bloc near its current ceiling, paired with a much larger but fragmented pool available to a rebuilt CDU campaign.

The government’s present weakness favors AfD. The September DeutschlandTrend found 15% satisfied with the coalition and 84% dissatisfied; Chancellor Friedrich Merz’s satisfaction rating was 13% (Infratest dimap). The economy has resumed modest growth—real GDP rose 0.3% quarter over quarter in the second quarter of 2026—but forecasts still show only moderate expansion through 2028, leaving neither a clear government recovery nor a fresh recession as the base case (Destatis; ifo Institute).

My regular-election models, built from the polling anchor and certified results, give 83% under no systematic drift, 75% with 25% reversion toward the 2025 result, and 64% with 50% reversion (DAWUM; Federal Returning Officer). I weight those cases 35%, 45% and 20%. I then include a low-single-digit discount for prohibition, party discontinuity or severe organizational failure: the February 26, 2026 Cologne court order slowed the intelligence-classification route, and a party ban would require a separate Federal Constitutional Court process (Cologne Administrative Court).

For election timing, I assign 86% to the regular 2029 window, 9% to an election during 2028 and 5% to an election by the end of 2027. Germany has held four early federal elections—1972, 1983, 2005 and 2025—while the current CDU/CSU–SPD coalition began with 328 of 630 seats and both partners now have strong electoral incentives to avoid an early vote (Federal Returning Officer; Bundestag seat distribution). The resulting timing-weighted calculation is:

$$0.86 \times 0.733 + 0.09 \times 0.782 + 0.05 \times 0.806 = 0.741$$

The three conditional terms are the estimated YES probabilities for a regular election, a 2028 election and an election by the end of 2027. Earlier elections score higher because there is less time for the current gap to close.

What's non-obvious

The resolution can be YES even if CDU/CSU combined finishes first. If CSU receives about five to six national percentage points, CDU alone overtakes AfD only when the combined Union leads AfD by more than that amount. An outcome such as AfD at 24% and CDU/CSU at 29% could therefore still leave AfD ahead of CDU individually (Federal Returning Officer historical results).

The second subtle point is that AfD does not need further growth. It can lose several points and remain first if the mainstream vote stays divided. The strongest NO path is not a sudden AfD collapse; it is a new CDU candidate recovering former Union voters while AfD slips into the low 20s.

Uncertainties

The long-horizon calibration is thin. Modern German polling supplies only two strong analogues at roughly the current distance from election day, and overlapping monthly poll changes are not independent observations (DAWUM archive). This makes model choice—especially the degree of mean reversion—more important than sampling error.

There was no national poll fielded after the September 6 Saxony-Anhalt result by the forecast cutoff, so its national bounce or backlash is unmeasured (DAWUM). CDU’s eventual leader and chancellor candidate are also unknown. Electoral law could still be amended, although the current second-vote system is the proper baseline, and I could not confirm an active federal AfD prohibition application; absence of confirmation is not proof that no preparatory work exists (current electoral law; Bundestag prohibition debate).