France · Next legislative election

577 races. One Assembly.

If the Assembly were dissolved today: a two-round Monte Carlo engine, transfers and withdrawals calibrated on the 1,646 runoffs of 2017-2024, anchored to the last direct legislative poll.

Last updated : 2026-07-11 Polling anchor : 2025-09-22
577 constituencies
2 rounds simulated
289 seats to govern

Estimated first-round vote intentions

RN and allies 37.1 %Left (NFP) 29.0 %Centre (Ensemble) 17.0 %LR and centre-right 12.0 %Far left 1.1 %Others and regionalists 3.5 %

Indirect estimate: last direct legislative poll October 2025 (5 anchor polls). Movement since then carried by presidential intentions (differential bridge), uncertainty ±3.2 pts.

The map, both rounds

Projected final winnerFavourite bloc in the central scenario (republican front at its 2024 level). Click a constituency for its page.
First round — leading blocBloc leading first-round intentions, before transfers and withdrawals.

The 11 overseas-French constituencies have no mappable territory: absent from the map, but present in the index and pages.

Three withdrawal scenarios

Republican front — 2024 level

Third-place withdrawals as in 2024: left 95%, centre 82%, right 28%.

45 %probability of an absolute RN majority

  • RN and allies 285 [231–340]
  • Left (NFP) 163 [123–203]
  • Centre (Ensemble) 68 [36–103]
  • LR and centre-right 48 [36–62]

Weakened front

Withdrawal probabilities halved — discipline erodes.

72 %probability of an absolute RN majority

  • RN and allies 310 [260–359]
  • Left (NFP) 157 [118–199]
  • Centre (Ensemble) 55 [28–85]
  • LR and centre-right 42 [32–54]

No front

Every qualifier stays in. Three-way runoffs everywhere.

90 %probability of an absolute RN majority

  • RN and allies 335 [289–379]
  • Left (NFP) 152 [113–193]
  • Centre (Ensemble) 42 [21–66]
  • LR and centre-right 37 [27–47]

Most contested constituencies

58 constituencies change winner depending on the front scenario. The 15 most uncertain (central scenario):

How this projection is built

  1. The problem: sparse polling

    French legislative polls are event-driven — they only appear during crises. The last direct wave dates to September-October 2025 (five polls, three firms). Nothing since.

  2. The differential bridge

    The level starts from that October 2025 anchor. The movement comes from 2027 presidential intentions, measured continuously: we keep only the change, applied one-for-one (β = 1, imposed for lack of co-observation windows, but validated on those available). Uncertainty is ±3.2 points and grows as the anchor ages.

  3. From votes to seats

    The national result is Monte-Carlo simulated across all 577 constituencies with the real rules: qualification at 12.5% of registered voters, direct election at 50% of votes cast plus 25% of registered. Vote transfers are estimated by duel type over the 1,646 runoffs of 2017-2024, then reconciled (Bradley-Terry). Withdrawals are measured — in 2024, third-place candidates stood down 95% on the left, 82% in the centre, 28% on the right.

  4. Three scenarios, not one certainty

    Because the future level of withdrawals is unknowable, we bracket it in three republican-front scenarios rather than freezing one. The backtests justify it: 2022 lands at 7.4 seats of mean in-sample error, 2024 places the far right within ±3 seats — but transfers drift sharply from one election to the next.

What this projection is — and is not

  • No direct legislative poll since October 2025: the level comes from that anchor, the movement from presidential polls (β=1, validated on the available co-observation windows).
  • The first round applies the real qualification threshold (12.5% of registered voters); the runoff simulates transfers and withdrawals calibrated on 2017-2024.
  • Future withdrawals are unknowable: each scenario is a conditional forecast, and together they define the range of plausible outcomes.
  • Turnout assumed at the 2024 level (66.7%). Lower turnout means fewer three-way runoffs and a different map.
  • Published backtests: 2022 (seat MAE 7.4 in-sample) and 2024 (far right within ±3 seats in-sample; left/centre residual documented).