A 2018 federal court order struck down a Kansas law that had already blocked more than 30,000 eligible American citizens from registering to vote, simply because they could not produce a passport, birth certificate, or naturalization paper on demand. Seven years later, Utah ran the same kind of citizenship check on its own voter rolls -- reviewing roughly 1.8 million active voters in 2025 -- and found exactly one noncitizen.^[3] Those two outcomes, both already closed, are the real record of what happens when a documentary-proof-of-citizenship requirement meets an actual voter file. The SAVE Act, passed twice by the U.S. House of Representatives, would apply that same standard to every federal election in the country.

What the bill actually does, and what already happened in the House

H.R. 22, the Safeguard American Voter Eligibility Act, passed the House 220-208 on April 10, 2025.^[1] A reintroduced, expanded companion measure -- carrying the same core requirement -- passed the House a second time, 218-213, on February 11, 2026.^[2] Both versions require a person to show in-person documentary proof of citizenship -- a passport, a birth certificate paired with a photo ID that matches their current legal name, or naturalization papers -- to register or re-register to vote in a federal election. Both also require states to remove noncitizens from their voter rolls and attach enforcement provisions, including potential criminal liability, for election officials who don't comply.^[2] As of this writing, the bill has stalled in the Senate: Republicans hold 53 seats, seven short of the 60 needed to clear a filibuster, and two attempts to attach it to budget reconciliation packages -- April 23 and June 4, 2026 -- both failed on 48-50 votes, with four Republicans joining every Democrat against it.^[6]

Five states already ran this experiment

The federal bill's fate in the Senate is still open. What is not open is what happened when states enacted the identical mechanism on their own. Kansas's 2018 experience is the clearest closed case: the same documentary-proof requirement, running for real against a real voter file, prevented more than 30,000 eligible citizens from registering before a federal court order struck the law down.^[3] New Hampshire enacted a documentary-proof law in 2024 and expanded it to mail registration in 2025; in that year's town elections, at least 96 voters were turned away for documentation problems, including one married woman who lacked the marriage license needed to prove her legal name change and had to return three separate times to register. A federal judge blocked enforcement of New Hampshire's law on May 28, 2026, citing constitutional violations.^[3] Utah's 2025 review is the sharpest illustration of the mismatch between the stated problem and the actual find: checking roughly 1.8 million active voters turned up one noncitizen.^[3] Going into the 2026 midterms, five states -- Arizona, New Hampshire, South Dakota, Utah, and Wyoming -- will require some form of documentary citizenship proof to register or update a registration.^[3]

Who the federal version would reach

The Brennan Center for Justice estimates that roughly 21 million American citizens lack ready access to the kind of documentary proof of citizenship the SAVE Act would require, with the burden falling disproportionately on people of color and younger voters. Roughly half of Americans do not hold a valid passport at all.^[4] A separate, specific gap sits inside that 21 million: a Center for American Progress analysis, reported by CBS News, estimates that as many as 69 million women and 4 million men currently hold a legal name that does not match the name on their birth certificate -- the product of a marriage-related name change, which roughly 84% of married women make. Under a documentary-proof requirement, a birth certificate and a current ID that don't share a name is exactly the kind of mismatch New Hampshire's 96 turned-away voters ran into in practice.^[5]

Why does this matter? The honest way to read the SAVE Act is not as a prediction about what a documentary-proof requirement might someday do -- it's as a bill that would nationalize a mechanism five states have already run, with a documented record already on file. Kansas blocked over 30,000 eligible citizens under the same requirement before a court intervened. New Hampshire is turning voters away in real town elections right now, for exactly the name-mismatch and missing-paperwork problems a birth-certificate-plus-current-ID requirement predictably produces. Utah checked 1.8 million voters against citizenship records built to catch the fraud this bill is written to prevent, and found one. That's a different mechanism than the ballot-screening, mail-delivery, and polling-place closures already documented on DoAyni -- a fourth pressure point on the same act of voting, this one at the registration desk rather than the ballot box. Whether the Senate ever takes it up is still undecided. How many eligible citizens the mechanism itself catches, and how many it doesn't, is no longer a question -- five states already ran the experiment, and the record is in.

Addendum: the "just get a passport" fix has a price tag

A passport is one of the documents the SAVE Act would accept as proof of citizenship. It is not free. Following the State Department's April 2026 fee update, a first-time adult passport book costs $165 total -- a $130 application fee plus a $35 execution fee -- or $65 for a passport card alone; renewing an existing passport by mail avoids the execution fee but still runs $130.^[7] Getting one also usually requires the exact document -- a certified birth certificate -- that a meaningful share of the 21 million citizens without ready proof of citizenship don't have either, which makes "just get a passport" a two-step cost for the people this bill would burden most, not a one-step fix.

Why this belongs here: the piece above treats the 21 million figure as a paperwork gap. It is also a straightforward cost -- $165 is real money to someone who was never going to travel internationally and had no other reason to own a passport, layered on top of the time and transportation cost of reaching a birth-certificate office and a passport acceptance facility in the first place. A requirement that can be satisfied by paying roughly $165 is not equally binding on everyone it applies to.

Addendum: "go through the state" is its own weeks-long process

For anyone whose birth hospital has since closed, the certified record itself doesn't disappear with the building -- state vital records offices, not hospitals, are the actual legal custodian of a birth certificate. But retrieving it means starting a separate process with the state where the birth happened, not the state where someone lives now, and that process has its own real cost in time and money: Illinois' current mail-request processing time is about 12 weeks; South Carolina's runs closer to 4; most states charge a separate $12 to $25 per certified copy on top of whatever the certificate is ultimately needed for.^[8][9] None of that is optional friction layered on by an especially disorganized applicant -- it is the standard process, for anyone who needs a replacement record and doesn't already have one in a drawer.

What this is actually weighed against: the SAVE Act's premise is that noncitizen voting is a large enough problem to justify this. The evidence in this piece says otherwise -- Utah's 2025 citizenship review of 1.8 million active voters found exactly one noncitizen. A separate, widely cited academic review found 31 credible instances of voter impersonation fraud, of any kind, out of more than one billion ballots cast nationally between 2000 and 2014.^[10] Voting is a constitutional right, not a discretionary privilege the state extends conditionally. A fix built to stop a problem this rare, that costs the people it burdens real weeks and real dollars to comply with, is not a neutral paperwork update -- measured against how rarely the harm it targets actually occurs, its practical effect is to make voting harder for millions of eligible citizens in order to stop a small handful of proven cases nationwide.

Addendum: the small number of proven cases don't fit a clean partisan story either

Among the roughly 170 voter-fraud defendants nationally with a known party affiliation, tracked across the 2016-2020 elections, the split is close to even -- 39.4% Democrat, 41.1% Republican.^[11] But the most heavily covered single cluster of 2020 cases came from one place: at least four residents of The Villages, Florida -- a retirement community and reliable Republican stronghold -- were charged and convicted of casting ballots in two states, several of them registered Republicans who owned homes in both.^[12] Separately, a Washington Post investigation of Republican-led election-integrity prosecution efforts found the opposite pattern in who actually got charged: defendants were disproportionately Black, Hispanic, and Democratic.^[13] None of this supports a clean story in either direction. What it does show is that the small number of real cases -- roughly split by party, headlined by a Republican cluster, prosecuted in a pattern that skewed toward minorities and Democrats when officials went looking -- doesn't match the specific, citizenship-focused threat the SAVE Act is built to stop.

Addendum: this is the same argument as the mail-ballot piece, not a separate one

This piece and the USPS ballot-screening and polling-place-closure piece are not two examples of a broader trend -- they are one argument, run on two different administrative levers. One adds a documentary-proof requirement at registration; the other adds a barcode-scan rejection rule, fewer rural mail trips, and 27,000 fewer polling places since 2012. Different mechanisms, different agencies, different stated justifications -- fraud prevention in one case, cost and logistics in the other. Both land on the same right, both add real, measurable friction (weeks of processing time, a document fee, a longer drive, a scan that can reject 10,000 ballots at once), and in both cases the friction is large relative to how well-documented the problem it addresses actually is. Measured together, not separately, the pattern is not "a series of unrelated administrative decisions." It is a consistent direction.

Addendum: "voting is state-run" is the textbook version, not the actual mechanics

The Constitution doesn't hand elections to the states without qualification, and it doesn't hand them to Congress either -- it splits the power deliberately. Article I, Section 4 gives state legislatures the first word on the "times, places and manner" of federal elections, then adds a federal override in the same sentence: "the Congress may at any time by Law make or alter such Regulations."^[14] That override was not an afterthought. Hamilton defended it directly in Federalist No. 59, arguing that without it, a state legislature could "annihilate" the federal government outright simply by refusing to hold elections for it -- every government, he wrote, "ought to contain in itself the means of its own preservation."^[15]

The most recent Supreme Court case to test this power directly cuts the opposite direction from what the SAVE Act would do, which is itself the point. In Arizona v. Inter Tribal Council of Arizona (2013), Arizona tried to require documentary proof of citizenship on top of the federal voter-registration form's sworn citizenship attestation. The Court ruled 7-2, in an opinion by Justice Scalia, that federal law already governing that form preempted Arizona's add-on -- a state cannot impose a stricter documentary standard than the one Congress set.^[16] That is real, exercised, and upheld federal power over how federal elections are registered -- just constraining a state from adding more than Congress required, not (yet) a ruling on whether Congress itself can mandate a stricter national standard the way the SAVE Act would. "The federal government has no business in state-run elections" isn't the accurate version of this fight. The accurate version is a real, textually explicit federal power whose actual limits -- what Congress can require, not just what states can add -- are the live question this bill would finally test.

Addendum: this is the argument the Civil War's resolution actually settled, and never finished settling

The war itself was fought over slavery, not voting -- but its direct legal resolution is exactly where this argument, applied specifically to voting, first got decided. The 15th Amendment, ratified February 3, 1870, states plainly that the right to vote "shall not be denied or abridged by the United States or by any State on account of race, color, or previous condition of servitude" -- and gives Congress the power to enforce it.^[17] That is a direct federal override of a state's control over its own voter qualifications, written into the Constitution specifically because the war's outcome required an answer to whether a state could use that control to functionally exclude a class of citizens from the vote.

The amendment did not end the argument; it just moved it. For nearly a century, states found workarounds that were race-neutral on paper -- poll taxes, literacy tests, grandfather clauses -- while achieving exactly what the 15th Amendment was written to stop. It took the Voting Rights Act of 1965, signed specifically "to enforce the fifteenth amendment," to give that promise real force: Section 5 required states and counties with a documented history of discrimination to get federal approval, called preclearance, before changing any voting rule at all.^[18] That enforcement mechanism itself didn't survive intact -- the Supreme Court's 2013 ruling in Shelby County v. Holder struck down the coverage formula that determined which jurisdictions needed preclearance, leaving Section 5 without a working enforcement list.^[19] The SAVE Act is not a new argument. It is the latest round of the one the country has been having, with real legal ground shifting underneath it, since Reconstruction: whether a state's control over the "manner" of its own elections can be used, in practice, to make a citizen's right to vote not count -- and how much power the federal government actually has to stop it when it happens.