Voting With Biometric Recognition: 3 Stages, Real Failures and Open Questions

Biometric voting system: ballot slip inside a glowing fingerprint pattern

A biometric voting system sounds like something from a thriller. You press your thumb on a pad and your vote is counted. Real systems work very differently, and the gap between the idea and the practice is where most of the debate lives.

In almost every country that uses one, voting with biometric recognition means something narrower. A fingerprint or a face scan proves who you are. The ballot itself stays separate, on paper or on a machine, and the biometric step only decides whether you may cast it.

That narrow job still carries big stakes. When it works, it can remove duplicate voters and block impersonation. When it fails, people can queue for hours or leave without voting. This guide follows the three stages where biometrics enter an election, shows what happened in recent votes and explains what the rules say.

What a biometric voting system actually checks

Election experts describe two kinds of check. Verification asks a simple question: are you who you say you are? The system compares your fingerprint or face with one stored record. Identification is harder because it searches a whole database to work out who you are. At the polling station the job is usually verification.

The same guide makes a point that is easy to miss. A match proves identity and nothing else. It cannot show that someone is a citizen or old enough to vote or registered in the right area. Those facts come from other records.

Biometrics enter an election at three points, and each one needs a plan for the day the technology fails. The same logic sits behind phone unlocks and bank logins. Our guide to biometric authentication walks through that process step by step.

Three stages where biometrics enter an election Registration, polling station verification and remote voting, each backed by a fallback such as a manual lookup. Where biometrics enter an election 1 Registration A kit captures fingerprints, a photo and sometimes an iris. Software then removes duplicate records. Goal: one voter, one record 2 Polling station A scanner compares the voter’s fingerprint or face with the stored record for that polling station. Goal: stop impersonation 3 Remote voting A phone selfie is matched to an ID photo before the ballot unlocks. Rare and heavily debated. Goal: reach absent voters Every stage needs a fallback A manual lookup, a photo check or a backup register, so no eligible voter is turned away.
Source: BiometricToday diagram based on International IDEA, Introducing Biometric Technology in Elections (2017).

Biometric voter registration: one voter, one record

Biometric voter registration is where most programs begin. Staff use a kit with a laptop, a camera and a fingerprint reader. Software turns each print into a template and compares it with every other record to catch duplicates.

International IDEA counted more than 50 countries using biometrics in elections as of 2016. Virtually none were in Europe, while about half of the countries in Africa and Latin America were. In its survey of more than 130 election bodies, 35% captured biometric data at registration.

Use of biometrics by election bodies, 2016 survey 35 percent captured biometric data at registration, 25 percent used biometric information at polling stations and 9 percent used an electronic biometric ID system at polling stations. Biometrics in elections: share of surveyed election bodies (2016) Capture biometric data when registering voters 35% Use some biometric information to identify voters at the polling station 25% Use an electronic biometric ID system at the polling station 9% Many of the 25% only check a photo on a printed voter list.
Source: International IDEA, Introducing Biometric Technology in Elections (2017), using its ICTs in Elections Database, September 2016. More than 130 election bodies surveyed.

Bangladesh is the best-known success story. A court ordered a clean-up of a voter list that held about 12.2 million wrong or duplicate names. The election commission then registered 80.5 million voters in 11 months, and an independent statistical review found the data accurate before the December 2008 vote. IDEA’s case study adds that observers later called that election the best in the country’s history.

Newer programs borrow from existing ID databases. Kenya’s electoral commission captures fingerprints, a facial image and an iris scan when it registers a voter and keeps one valid record per person. Nepal let citizens with a national ID register online ahead of its March 2026 election because the ID agency shared its biometric data with the election commission.

Reuse saves money. IDEA puts the start-up cost of biometric registration at 2 to 10 euros per voter and says the bill falls when data comes from an existing register. Our roundup of the uses of biometrics shows how the same tools work at borders, banks and airports.

Biometric voter verification: the scanner on election day

On polling day biometric voter verification takes over. A device compares a voter’s fingerprint or face with the register for that polling station. A clean match clears the voter to receive a ballot. Ghana’s election commission stressed that its devices only verify voters and are never used to cast a vote.

Ghana bought 74,800 new devices for its December 2024 election. They carry both fingerprint and facial checks, and the commission’s chair said voters who could not pass the biometric check would be verified manually.

Uganda took a similar path for its January 2026 election. Its Electoral Commission said it had acquired 109,142 kits for 50,739 polling stations. The kits match fingerprints first and fall back to the face when a print will not read. They also turn away anyone who arrives at the wrong station.

Why use a face as the backup? IDEA’s guide says face matching typically fails more often than fingerprint matching. Iris scanning is more reliable than both, but the guide described it as costly and rarely used in elections. Our comparison of face recognition and fingerprint systems explains the trade-offs.

Mongolia offers a calmer example. Its offline laptops show each matched voter’s photo on a large screen so observers can see it too. In 2016 about 70,000 of the 1.4 million identified voters, roughly 5%, needed a manual lookup because a print would not verify or their data was missing from the laptop.

When the scanner fails: Uganda, Nigeria and Ghana

Uganda’s January 2026 vote showed the risk. The Uganda Human Rights Commission observed 990 polling stations in 78 districts and said the kits worked effectively at only a limited number of them. Some stations in Yumbe and Wakiso never received kits. At about 9:30 a.m. the commission’s chair told officials to switch to the manual voters’ register, and polling was extended by an hour.

Officials and critics read the failure differently. The Electoral Commission said the machines worked when operators, credentials and logistics were in place. President Museveni said staff biodata had not been uploaded when polls opened. Critics argued that falling back to the manual register weakened the transparency the kits were meant to add. The human rights commission urged full testing of election technology before polling day.

Nigeria faced a similar test in 2023 with its BVAS devices, which check fingerprint and face. The Commonwealth Observer Group’s final report found that failures of the devices and the results portal undermined the credibility of the vote. It named too little testing and limited staff training as likely causes and urged offline backup options. The government has maintained that BVAS helped deliver what it called the most credible elections in its history.

Nigeria card reader test: success rate by state In a mock accreditation test before the 2015 elections, successful voter accreditation reached up to 92 percent in the best state, about 36 percent in another state and 3.4 percent in the weakest. Same card readers, very different results Nigeria’s mock accreditation test before the 2015 elections: success rate by state Best-performing state up to 92% Another state about 36% Weakest state 3.4%
Source: International IDEA, Introducing Biometric Technology in Elections (2017), Nigeria case study. Results reflect fingerprint authentication of voters in a mock exercise, not election day.

Ghana’s 2024 vote ran in the middle. Some polling stations reported malfunctioning devices, and officials used face verification to clear the affected voters. The commission later said the devices had performed effectively and cut processing time.

…manipulation and malpractice can never be prevented by technology alone.

International IDEA, Introducing Biometric Technology in Elections

The patterns repeat across these cases. The table below pairs each common failure with what the published guidance suggests.

AreaWhat goes wrongWhat the guidance suggests
Fingerprint qualityWorn, damaged, calloused or dirty fingers fail to matchCapture more than one finger and keep a fallback
Face photosPoor lighting and old ID photos lower accuracyUse the face as a backup to fingerprints, not the only check
TestingToo little field testing and tight timelinesPilot in smaller elections and start 18 to 24 months ahead
Staff and set-upUntrained operators and missing log-in credentialsTrain staff and polling agents and test the full chain
Voters at the doorA failed scan can mean a long queue or a turned-away voterKeep a manual register so no eligible voter is excluded
Sources: International IDEA (2017), Commonwealth Observer Group (2023), Uganda Human Rights Commission (2026), Telangana election officials (2020).

Voting with biometric recognition on a phone

Remote voting is where the idea becomes most contested. The best-known case is Voatz, a US mobile app. Voters confirmed their identity with a photo ID and a selfie checked by facial recognition. West Virginia offered the app to overseas military voters in 2018 and later to voters with disabilities.

Security researchers at MIT reverse engineered the app in 2020 and reported flaws that could let attackers alter, stop or expose a vote. They judged that exploitation was within reach of a nation-state attacker. West Virginia and one county in Washington dropped their plans for the 2020 primaries. The company has said an independent federally certified test lab found its platform compliant with US requirements. The same report quotes a Denver elections director saying facial recognition did not work for every voter.

Estonia has voted online since 2005, and the report puts the online share of recent votes above 45%. Voters prove who they are with a state eID card, not their face. A report ordered by the country’s Information System Authority said facial recognition could work as a remote check where no secure eID exists. It still found the risks too high for now: false matches of up to 3% in poor conditions, privacy exposure and unequal access to good devices. In the report’s words, “Elections are by far one of the worst test grounds for any unrefined technology.”

Privacy, sensitive data and the law

A voter register that holds fingerprints is a high-value target. Unlike a password a fingerprint cannot be changed. IDEA warns that leaked original images carry a risk of identity theft that reaches well beyond elections. It calls for laws that limit who can see the data and why. Our explainer on biometric data covers how this information is stored and protected.

Europe’s data protection body has set a high bar. In September 2024 the Council of Europe published guidelines on voter registration and authentication under Convention 108. They treat biometric data as a special category. Such systems should be introduced only if older forms of identification have been shown to be inadequate, according to the text. The guidelines also say biometric data should never be shared with political parties.

India shows the debate in motion. Telangana tested a face recognition app at 10 polling stations in January 2020. Officials said a failed match would not stop anyone from voting and that photos would be deleted. They later put accuracy at about 80% and blamed poor lighting and old ID photos, while an opposition party objected on privacy grounds. In April 2026 the Supreme Court issued notice on a petition seeking mandatory fingerprint, iris and face verification of voters. The court noted that the change would need new rules and a large budget.

The case for and the case against

Supporters and critics often look at the same evidence and reach different conclusions. The table sets out the arguments each side makes, area by area.

AreaWhat supporters sayWhat critics say
FraudBiometrics sharply cut duplicate registration and multiple votingTechnology alone cannot stop manipulation, and exceptions to enrolment can be abused
TrustA clean register can rebuild confidence after a dispute, as in BangladeshElection-day failures can damage trust, as observers found in Nigeria and Uganda
CostReusing national ID data lowers the bill, and kits can serve several elections2 to 10 euros per voter at the start, plus upkeep, licences and vendor lock-in
AccessVoter cards from registration can become widely accepted ID, as in FijiQueues, unreadable prints and unequal phone access can exclude real voters
DataStored templates are often proprietary, which limits their use if leakedA breach of original images cannot be undone and affects people beyond elections
Sources: International IDEA (2017), Electoral Commission of Uganda (2026), Council of Europe (2024).

IDEA’s own conclusions sit between the two camps. It calls biometrics very efficient at reducing multiple registration and multiple voting. It also says the added value is likely limited where citizens already hold trusted ID and fraud is a minor problem.

Biometric voting system FAQ

Can you vote with your fingerprint?

Not in the way the phrase suggests. Fingerprint voting normally means a fingerprint check before you are handed a ballot. The vote itself is cast on paper or a separate machine.

What happens if the scanner cannot read my fingerprint?

Well-run systems have a fallback. Mongolia looks the voter up by hand and shows the photo on a large screen. Ghana allows manual verification and has used face checks. Uganda told officials to use the manual register where kits failed.

Is there a biometric voting machine?

The devices at polling stations are usually electronic poll books or handheld verification kits. They check who you are. They do not record your choice.

Which countries use a biometric voting system?

International IDEA counted more than 50 countries using biometrics in elections as of 2016, mostly in Africa and Latin America. Recent examples include Bangladesh, Ghana, Kenya, Nigeria, Uganda and Nepal. Each uses the technology for different stages.

Does a biometric check put ballot secrecy at risk?

IDEA lists that belief among the rumors election bodies should address with clear voter information. The check confirms identity before voting. It is a separate step from marking a ballot.

Questions worth asking before the next election

  • Should a failed scan ever slow a voter down, or should the manual register always be the safety net?
  • Who should hold a nation’s fingerprint data: the election body or the ID agency?
  • Is a face check on a phone a fair price for reaching voters who cannot get to a polling station?
  • What would convince you that a system had been tested well enough before polling day?

Where do you see the line between a safeguard and an obstacle? Tell us in the comments.

Sources

⚠ Disclaimer: We work to keep this content accurate, but details may be incomplete or outdated, so please verify important facts with official sources. Any inaccuracy about the news, a company, a country or anything else is unintentional. Please contact us immediately if any such case happens.

We will be happy to hear your thoughts

Leave a reply

Biometric Today
Logo