How this report was built

The totals of each identity’s first line are official. The parts are counted from our digitised records. Where our records do not cover the whole official count, the report shows the difference as “Not in our data”.

1. Official counts. The ECI’s assembly-wise figures give, for each of the 294 assemblies, the electors under adjudication, those found eligible and those found not eligible. They were provided by the ECI in reply to an RTI application by Shri Isha Khan Choudhury, Member of Parliament, Dakshin Malda, dated 17/08/2026, vide memo no. 8551-Home(Elec). They are the base of every percentage for the state, districts and assemblies.
2. Deleted voters. The ECI published the electors found not eligible in supplementary lists per assembly and booth, as image-based PDFs. Preliminary digitisation of these lists was done by Sabar Institute. Civic Research Group verified those EPIC numbers, corrected wrong EPICs and added EPICs that were missing.
3. Included voters. The supplementary lists of electors found eligible and included in the e-roll after adjudication were digitised entirely by Civic Research Group: name, EPIC number, gender, age and booth.
4. Appeal status. Each EPIC was checked on the Tribunal appeal portal, between 5 September and 20 October 2026. For deleted voters the portal gives the appeal status, appeal number, dates and, where decided, the Tribunal decision; for appeals checked before 18 September 2026 that decision showed whether the Tribunal approved or rejected the appeal, and after the ECI made a one-time password (OTP) mandatory it no longer does. For included voters the portal shows whether an appeal against inclusion exists and whether it is pending or disposed.
5. Invalid EPICs. EPICs that did not resolve on the portal went through a further portal check. Those still invalid were checked by hand against the ECI’s lists. This work was done by unpaid volunteers; the project received no external funding from any source, institution or organisation.
6. Booths and local bodies. Booths are matched to municipalities, wards and gram panchayats using the Form 20 booth list, with local body names checked against the Local Government Directory.
7. Aggregation. Booth, local body, assembly, district and state figures are added up from the voter records. Nothing is estimated or modelled.

Under adjudication, included, deleted

The ECI’s official counts, per assembly, provided in reply to an RTI application: electors under adjudication (60,06,675), found eligible and included in the e-roll after adjudication (32,77,360), and found not eligible and deleted (27,28,500). “Deleted” in this report always means the official count found not eligible. Source.

In our data

Voters in the lists we digitised and checked on the appeal portal: 27,06,057 deleted voters and 32,53,401 voters included after adjudication.

Coverage

Voters in our data as a share of the official count for the same place. Coverage below 100% means some names could not be read or matched; above 100% means the published lists hold more names than the official count. The difference is shown as “Not in our data”.

Applied to Tribunal

Deleted voters who appealed against their exclusion: 22,17,619 (81.3% of deleted). It always equals the appeals pending at the Tribunal plus those disposed by the Tribunal (21,73,997 + 43,622). By how the appeal was made, they are counted in three ways:

• Submitted online: the application was made online. 17,67,095 (64.8% of deleted).
• Submitted offline: the application was made offline, on paper. 1,48,846 (5.5% of deleted).
• Applied, channel unknown?: the voter applied, but the channel is not known. 3,01,678 (11.1% of deleted).

Not applied

Deleted voters with no appeal on record: 4,88,438 (17.9% of deleted).

Pending and disposed

Pending: appeals the Tribunal has not decided yet (21,73,997 against exclusion, 15,26,933 against inclusion). Disposed: appeals the Tribunal has decided. Against exclusion, 30,828 approved, 6,444 rejected and 6,350 disposed with the outcome unknown?. Against inclusion, 81,242 disposed, all with the outcome unknown.

Appeals against inclusion

Appeals filed by the ECI or a third party to remove a voter whom a judicial officer had found eligible and who was included in the e-roll: 16,08,175 in our data. These voters stayed on the roll and voted in the 2026 assembly election. Before 18 September 2026 Civic Research Group checked a few appeals against inclusion; in almost all of them the appeal had been submitted by the “Addl CEO” (Additional Chief Electoral Officer). After 18 September the ECI’s mandatory OTP stripped that detail from the portal and made independent verification very hard.

Base of percentages

For the state, districts and assemblies, shares of deleted or included voters use the official counts. Booths and local bodies have no official count, so they use the voters in our data.

Local body numbers

A municipality, ward or gram panchayat adds up the booths the Form 20 booth list places in it. A corporation or municipality includes all its wards.

• Lists that machines cannot read. The ECI published the roll and its lists of included and excluded electors as PDFs with embedded images. Converting them was difficult and some errors were unavoidable, so a small number of EPICs could not be resolved.
• No official appeal data. The ECI publishes no consolidated or assembly-wise appeal status report, so the status had to be checked one EPIC at a time on the Tribunal appeal portal.
• One-time password from 18 September. Civic Research Group checked individual EPICs on the Tribunal appeal portal between 5 September and 20 October 2026. For appeals checked before 18 September 2026 the portal showed the Tribunal’s outcome (approved by Tribunal or rejected by Tribunal). After the ECI made a one-time password (OTP) mandatory for checking an individual EPIC’s appeal status, that detail is no longer accessible to us, so the report shows such appeals as disposed by the Tribunal with the outcome unknown (“Disposed by Tribunal (outcome unknown)”). The OTP makes independent large-scale checking much harder, and figures gathered before the change are hard to refresh.
• Who filed the appeals against inclusion. Before 18 September 2026 Civic Research Group checked a few appeals against inclusion; in almost all of them the appeal had been submitted by the “Addl CEO” (Additional Chief Electoral Officer). After 18 September the ECI’s mandatory OTP stripped that detail from the portal and made independent verification very hard.
• Disposed by Tribunal (outcome unknown). ?For 6,350 appeals against exclusion, the portal response we could obtain shows that the Tribunal took a decision, but not whether it approved or rejected the appeal: these were checked after the OTP became mandatory, or the portal never showed the outcome. These are never counted as approved, rejected or “other”.
• Applied, channel unknown. ?For 3,01,678 deleted voters the response shows that they applied to the Tribunal, but not whether the application was made online or offline.

• Deleted voters. Our data holds 27,06,057 of the 27,28,500 electors officially found not eligible (99.18%).
• Included voters. Our data holds 32,53,401 of the 32,77,360 electors officially found eligible (99.27%).
• Official figures that do not add up. In 43 assemblies, eligible plus not eligible differs from the total under adjudication; statewide the difference is 815. The figures are shown as given. See the “Official and coverage” columns of the assembly table.
• EPIC numbers. 27,06,057 deleted-voter records have a valid EPIC number.
• Coverage above 100%. In 2 assemblies our records hold more names than the ECI’s official count on at least one side; 0 of them are among the assemblies whose official figures do not add up. The published lists are the source of our records, so the official split between eligible and not eligible looks incomplete there. Coverage is shown as it is, never capped at 100%.
• EPICs in both lists. 244 EPIC numbers appear in both the included and the deleted list of the same assembly. Each list is counted as published, so these electors are counted on both sides.
• Records with no booth number. 33 deleted-voter records in 1 assembly have no booth number, so they count toward their assembly but no booth or local body: AC 166 (33).
• Local body names. The Form 20 booth list names the municipality, ward or gram panchayat of every booth in 249 of 294 assemblies. The other 45 assemblies have booth data but no local body names, so their voters do not appear under any local body.
• Gender. Gender is counted from the labels in the lists. 2,335 deleted-voter records have none and are shown as “Not recorded”. The labels say nothing about any person’s gender, eligibility or legal standing.

Assemblies without local body names (45)

Underlying data, public repositorygithub.com/civicresearch/wbsir

Others are welcome to independently use, verify and check the data, and to use it for reporting, with proper citation and a link to https://appeal.civicresearch.in/.

Civic Research Group, West Bengal SIR 2026 Adjudication & Tribunal Appeals Report, https://appeal.civicresearch.in/

Please credit Civic Research Group and link to appeal.civicresearch.in when you use these numbers. Reuse terms and contact.