How the BJP won Assam in 2026: what the polls said and what happened

Voting was on 9 April 2026 and counting on 4 May. The NDA, made up of the BJP, the Asom Gana Parishad and the Bodoland People’s Front, won 102 seats. The opposition front, the Asom Sonmilito Morcha, won 21 by Wikipedia’s count. That is Congress with 19 and the Raijor Dal with 2. Outside both camps, the AIUDF won 2 and the Trinamool Congress won 1. Himanta Biswa Sarma was sworn in for a second term as Chief Minister on 12 May.

Three things stand out. The BJP won 82 of the 90 seats it contested, by our data. Congress’s vote share barely moved, yet its seats fell by ten. And the map itself was different: Assam’s 126 seats were redrawn in 2023, so 2026 was the first Assembly election on the new boundaries. We look at what the numbers can and cannot tell us about each of these.

The one pre-poll survey we could confirm is an IANS-Matrize poll, dated 15 March 2026, as listed on Wikipedia. It gave the NDA 96 to 98 seats and the Congress-led front 26 to 28, with vote shares of 43% to 44% and 39% to 40%. It was an opinion poll taken about three and a half weeks before voting, and it is one poll, not a consensus of all polls.

So the poll got the winner right and understated the margin: the NDA finished above its range on seats and votes, and the opposition front finished below its range on both. The NDA figure of about 48.0% is our own sum of the BJP, AGP and BPF votes over all votes polled, and it matches the 48.01% on Wikipedia. The 35.95% for the opposition front is Wikipedia’s figure, which we have not recomputed.

Earlier results also pointed the same way, though not strongly. In the 2024 Lok Sabha election in Assam, the BJP won 9 of the state’s seats and Congress 3, and the two parties polled almost the same share of the vote. Assembly and Lok Sabha shares are not directly comparable, but the table shows that the BJP’s share in 2026 was its highest of the four elections and Congress’s was not.

Exit polls, taken from people leaving polling booths, were reported on 30 April, before counting on 4 May. Every one we found expected the NDA to win. Most gave it between the high 80s and around 100 seats. One, People’s Pulse, was well below the rest. The table below is based on The Federal and The Tribune, which report the same figures.

The result, 102, was above the top of every range in the table. The Congress-led front’s 21 seats was inside only one range (Poll Diary’s 15 to 26) and below the bottom of the other six. Wikipedia’s summary gives an exit-poll average of about 49.7% for the NDA and 38.3% for the opposition front, against about 48.0% and 35.95% in the result. The two newspapers we used give no vote shares.

Wikipedia’s summary describes the NDA’s exit-poll range as 82 to 101 seats, which does not match People’s Pulse’s 68 to 72 in the two newspaper reports. We use the agency-by-agency figures from the newspapers. We have not judged the quality of any poll’s method.

Two of these lines need a note. In 2021 the BPF was in the Congress-led alliance (called Mahajot then) and the UPPL was with the NDA, according to Wikipedia. By 2026 the BPF had moved to the NDA, and The Week’s results page says the UPPL left over seat-sharing in March 2026. So the NDA’s and the opposition’s seat and vote totals in the two years are not made of the same parties. Wikipedia gives the 2021 NDA as 75 seats with 44.51% of the vote and the 2026 NDA as 102 with 48.01%. Our own sum for the BJP, AGP and BPF in 2026 is about 48.0%.

The BJP’s vote share rose by about 4.6 points, and Congress’s rose by about 0.2 point. The AIUDF’s fell by about 3.8 points. Congress’s seats fell because of where its votes were, not how many it had, an issue we return to below.

Turnout was about 86%, against about 82% in 2021. Votes cast rose from about 192.2 lakh to about 216.6 lakh, which is 24.4 lakh more, or about 13%. The electoral roll in our data grew by about 7%. Where sources give different turnout figures (85.4% on polling day, 85.96% and 86.25% in later compilations), we say “about 86%”.

The BJP added about 18.1 lakh votes, which is roughly three-quarters of the 24.4 lakh extra votes cast in the state. Congress added about 7.6 lakh. The AIUDF lost about 6.0 lakh, the AGP about 1.2 lakh and the UPPL about 1.6 lakh. Independents polled about 1.9 lakh fewer. The BPF gained about 1.6 lakh, and as noted above it changed sides.

The journalist Sandipan Talukdar, quoted in Outlook, ties the changes in border-region seats to minority voters backing Congress after the AIUDF was left outside the opposition front. The totals here fit that account, since Congress gained and the AIUDF lost by similar amounts. But they cannot prove it.

Comparing vote counts is arithmetic, not proof of voters switching. A voter who did not vote in 2021 and voted BJP in 2026 looks identical, in these totals, to a voter who switched from another party. The rise in turnout means new or returning voters could account for part of any party’s gain.

Comparing seats between 2021 and 2026 is harder in Assam than usual. The 2023 redraw changed names and boundaries, and only 73 of the 126 seats carry the same name in our data for both elections. Even those may cover different ground. So the table below is a comparison of who finished first and second, which we can do for every seat. After it, we list the same-name seats that changed hands.

The BJP and Congress finished first and second in 65 seats in 2021 and 64 in 2026. In 2021 the BJP won 51 of them and Congress 14. In 2026 the BJP won 62 and Congress 2. Of Congress’s 19 wins in 2026, 17 came in seats where the runner-up was the AIUDF (10), the AGP (6) or the BPF (1), not the BJP. The three biggest winning margins in the state were all Congress wins: Gauripur by 119,097, Jaleshwar by 109,688 and Samaguri by 108,310, each over the AIUDF or BJP.

That is 18 of the 73 same-name seats. The other 55 were won by the same party both times. The 53 seats with new names or boundaries are not in this table.

The victories were mostly wide. The median winning margin across the state was 43,856 votes, against 23,763 in 2021. The BJP’s 82 winners had a median margin of about 45,600. It won 75 seats by more than 20,000 votes and 32 by more than 50,000, and none by under 5,000. Congress’s 19 winners had a median margin of about 66,300. Only two seats were decided by fewer than 5,000 votes, and both went to the AGP over Congress: Hajo-Sualkuchi by 724 and Bilasipara by 2,694.

Some individual seats have their own stories. Himanta Biswa Sarma won Jalukbari with 127,151 votes, by 89,434 over Congress’s Bidisha Neog. Congress’s Gaurav Gogoi lost Jorhat to the BJP’s Hitendra Nath Goswami by 23,182. The AIUDF’s Badruddin Ajmal won Binnakandi by 35,380. Akhil Gogoi of the Raijor Dal won Sibsagar by 17,272 over the BJP’s Kushal Dowari. Congress’s Debabrata Saikia lost Nazira by 46,701. Outlook reports that Congress won only one seat in upper and northern Assam, a claim we have not checked seat by seat.

The 2023 delimitation kept 126 seats but redrew them. Newsonair’s report of the Election Commission’s announcement lists 9 seats reserved for Scheduled Castes and 19 for Scheduled Tribes for 2026. See our explainer on delimitation and on reserved seats for how this works.

How much it mattered is disputed. Al Jazeera’s Arshad Ahmed, writing on 1 April, reported opposition leaders and analysts saying the redraw cut the number of Muslim-majority seats from about 35 to about 20, and that the Barak Valley lost two seats, going from 15 to 13. They describe this as communal gerrymandering. Al Jazeera quotes the BJP spokesperson Kishore Kr Upadhya saying the remapping was not communal and putting responsibility on the Election Commission. The Chief Election Commissioner had not responded to its questions.

Monir Hussain, writing in India Today NE on 6 May, puts the fall in Muslim-majority seats at about 35 to about 22 and argues it was one of three structural forces behind the result, along with welfare and the BJP’s organisation. Akhil Ranjan Dutta, a Gauhati University political scientist quoted in Outlook, reads the same change differently. He says the BJP presented delimitation as protecting the state’s indigenous people, and notes that Scheduled Tribe seats rose from 16 to 19.

The two estimates of Muslim-majority seats (about 20 and about 22) differ, and neither source shows its method. We have not counted them ourselves, since our data has no seat-level population figures.

Nothing official explains why voters chose as they did. What follows is what named commentators argued, in the order of their pieces. These are their views, not our findings, and we have not found a post-poll survey with breakdowns by gender, age, community or region that could test them.

Welfare is the theme they share most. Outlook’s 12 April piece by Fozia Yasin says the Orunodoi scheme pays Rs 1,250 a month to about 40 lakh women, and that on 10 March the state transferred Rs 9,000 to each of four million women, about Rs 3,600 crore in all. It also reports that Congress promised its own benefits, including health insurance of Rs 25 lakh and pensions for seniors. Whether those payments changed anyone’s vote is a claim, not a finding. We could not find survey data on it.

The BJP and its allies also ran on other themes. The Week’s results page says Sarma campaigned on action against infiltration, implementation of peace accords and welfare, while Congress under Gaurav Gogoi campaigned on anti-incumbency, jobs, prices and governance. Dutta describes the BJP campaign as also using language about illegal immigration and promising a Uniform Civil Code. We have not tested how far any of this moved voters.

The numbers add one caution. Congress’s vote share was almost flat, at 29.7% and then 29.8%, so its loss of ten seats cannot be a simple story of it losing voters. Under first-past-the-post, a party can hold its share and still lose seats if its votes are spread differently across seats, as our explainer shows. Hussain’s reading is that Congress could not turn its votes into seats in the redrawn constituencies. We cannot test that with seat-level data across two maps.

Wikipedia’s page on the 16th Assembly lists Ranjeet Kumar Dass of the BJP, who won Bhowanipur-Sorbhog by 39,152 votes, as Speaker, and Wazed Ali Choudhury of Congress as Leader of the Opposition. To see how a Chief Minister is chosen after a result like this, read our explainer.

• Who voted for whom. We could not find a post-poll survey for this election, so every explanation above rests on commentary and the results alone.
• Whether the extra 24 lakh votes were new voters, returning non-voters or switchers.
• How much the 2023 redraw changed the outcome. The two published estimates of Muslim-majority seats differ, and no one has shown what the 2026 result would have been on the old map.
• Whether welfare payments changed votes. Analysts argue they did. We found no data that tests it.
• Why the exit polls fell short of the NDA’s 102. Every range topped out at 101 or lower, and the result was 102.
• Whether changes to the voter list affected the result. We found no credible analysis linking the two, so we make no claim.
• Whether other pre-poll surveys exist. We list only the one we could confirm.
• The status of any court challenge to individual seats. We have not confirmed any, so we list none.

Seats, votes, margins and same-name seat changes are our own analysis of results published by the Election Commission of India, the same data behind the Assam 2026 election page and the 2021 page. Vote share is a party’s votes over all votes polled, including NOTA (2.67 lakh in 2026, against 2.20 lakh in 2021). That matches Wikipedia’s figures for the BJP (37.81%) and Congress (29.84%). Some sites divide by a total without NOTA and get slightly higher numbers, such as 38.3% for the BJP.

Where sources differ we say “about”. Turnout is reported as 85.38% on polling day and 85.96% or 86.25% later, and 82.05% or 82.42% for 2021. Our roll of about 2.51 crore electors is slightly above the 24,958,139 that Wikipedia lists. Congress’s tally is 19 in our data, on Wikipedia and in India Today NE, while Outlook’s post-result piece says 22, probably from early counting. Exit-poll and survey numbers are from the outlets named, and analysts are credited by name. If you find a number that does not match the Commission’s, tell us on the contact page.