Methodology & sources

One source, and only one

Every figure on this site comes from a single publication: the Visa Bulletin, issued once a month by the U.S. Department of State's Bureau of Consular Affairs. It is a table of cut-off dates — one per visa category and country of chargeability — announcing how far down each queue the State Department has reached. Nothing here is bought from a vendor, scraped from another website, crowdsourced, or inferred from case reports. If State did not print it, it is not on this site.

The Visa Bulletin is prepared by federal employees in the course of their official duties, so under 17 U.S.C. §105 it carries no copyright and is in the public domain. Republishing it needs no licence, no key and nobody's permission. Attribution to the State Department is customary, and this site gives it on every page. This site is not affiliated with, endorsed by, or connected to the U.S. Department of State or any government agency.

Two of the bulletin's charts are used: Final Action Dates and Dates for Filing, across the family-sponsored and employment-based preference categories. The Diversity Visa rank cut-off tables are printed in the same document and are not used — they are a different mechanism with different units, and mixing them into a priority-date tool would mislead. They are left out rather than half-supported.

How the bulletins are read — and why not from travel.state.gov

This is the part of the method most worth knowing, because it is unusual and a reader deserves to hear it from us rather than discover it. This site does not read the State Department's website directly. It reads the Internet Archive's captures of it.

There are two reasons. The first is that we cannot read the live page: requesting a bulletin from travel.state.gov with an automated client returns HTTP 403 from a Cloudflare bot gate. The published PDFs are behind the same gate. The second is that even without the gate it would not be enough: the live site reliably serves only the current fiscal year, so the 291-bulletin history on this site is not reproducible from it at all.

The Internet Archive solves both. Its replay endpoint returns the State Department's own originally-served bytes — the response as State sent it, not a re-rendering — so the figures here are still State's figures, read from State's own HTML. The Archive is the courier, not the author, and it asserts no rights over archived federal content. Every bulletin is fetched once, cached, and parsed offline; the crawler is throttled because that is the courtesy the Archive asks of automated clients.

The cost of this arrangement is a ceiling on how fresh the site can be, and it is set out in full below. We would rather read the live source. We cannot, so we say so.

What the archive covers

291 bulletins, from December 2001 to July 2026, carrying 29,443 published cut-offs. Of those cells, 16,454 carry a priority date, 12,323 say Current and 666 say Unavailable. Every one is stored alongside the exact text State printed in that cell, so any figure on this site can be traced back to the bulletin it came from without taking our word for anything.

December 2001 is a real floor, not a stopping point we chose. The Archive was probed for earlier bulletins under the same URL scheme — November 2001, October 2001, September 2001, July 2000 and July 1999 — and every one returns nothing. State's re-published historical bulletins begin in December 2001, so that is where this dataset honestly begins.

The bulletin's own shape changed partway through. For the first 161 months of the record there was one cut-off chart per preference system, and it is the Final Action Dates chart. The Dates for Filing chart was introduced in October 2015 and covers the 130 months since. That is why histories on this site come in two different lengths: it is the source's shape, not missing data, and no Dates for Filing value is ever invented for a month that never had one.

5 months are missing, and here they are

Five bulletins in that span cannot be read, because no capture of them exists — not in the Archive's index, and not from an independent per-month query either. They are absent from the public record we can reach, so they are absent here:

• March 2009 — no capture exists in the Wayback index
• September 2009 — no capture exists in the Wayback index
• October 2009 — no capture exists in the Wayback index
• November 2009 — no capture exists in the Wayback index
• October 2012 — no capture exists in the Wayback index

They are never filled in. Not from the month before, not from the month after, not by averaging the two. A missing month is a fact about the source, and inventing a cut-off for it would be inventing the one thing this site exists to report accurately. On a history chart the line simply breaks at a gap rather than drawing across it, because a line drawn across a hole is a claim about data that does not exist.

The gaps have a knock-on effect that is handled the same way. Where a category's cut-off changed across a missing month, the change is real but the interval is not one month — it is two, three or four. 200 movements in this corpus span more than one month for exactly this reason, and each one records how long it actually spanned and which months are missing from it. None is ever presented as a single month's movement.

How current this is — and the ceiling on it

The State Department publishes one bulletin a month, around the middle of the month before the one it governs. Past bulletins are immutable once published, so refreshing this site is not a merge but a full rebuild: the whole corpus is re-parsed from cache and only the new bulletin is fetched. That happens monthly.

But this site cannot be fresher than the Internet Archive. Because the live State page is closed to us, a bulletin does not exist for this site until the Archive has crawled it. That is a real limitation on the one day that matters most — the day the new bulletin drops — so here is the measured record rather than a reassurance. These are the last 6 bulletins and the day the Archive first captured each:

Across the last 24 bulletins (August 2024 to July 2026), 24 reached the Archive before the month they govern had even begun — a median of 19 days of headroom. The tightest was November 2024: first captured on 31 October 2024, 1 day before it took effect.

What that table does not tell youIt does not measure how far behind the State Department we are, and neither does anything else on this page — because State's publication moment is not observable. The Archive cannot capture a page before it exists, so a first capture is an upper bound on publication and nothing more. We will not turn that into a number. What we will say plainly is the consequence: on the day State posts a new bulletin, this site may still be showing the previous one, and it will keep showing it until the Archive crawls the new one. A site reading the live page can publish within minutes. We cannot, and the record above is the whole of what we know about the difference.

Two things bound the risk, and are worth stating for the same reason the risk is. Every figure carries the bulletin month it came from, on every page, so a reader is never left guessing which bulletin they are looking at — if this site is behind, it says which month it is showing rather than looking current. And the refresh asks the Archive what exists rather than asking the calendar, so a new bulletin is picked up whenever it lands, not on a schedule that could quietly miss it.

What the source itself gets wrong, and what this site does about it

The Visa Bulletin is a hand-maintained government document with a 24-year history, and it reads like one. The interesting failure is not that it contains mistakes — it is that the obvious way to read it turns those mistakes into silence. A parser that matches column headers on their correct spelling does not crash on a typo. It just returns a chart with a column missing, and nothing anywhere says so.

These are real strings from State's own HTML:

The first four are the same column — the one State prints as "All Chargeability Areas Except Those Listed", which this site calls All other countries and which is the most-read number on the whole chart. Nine bulletins spell its header wrong one way or another. Note the third: the misspelling is hose, not in the word a matcher would key on — so even a rule keyed on the word "Those" would have missed it. Matching is therefore deliberately fuzzy: the header is squashed to letters and tested for chargeab or chargab, so all nine are read. The result is checked on every build, and it is the reason this section is not just a story: the All other countries column is present in 291 of 291 bulletins — every single one. If a spelling ever defeats the matcher, that count drops and the build stops.

The rule behind all of it: report, never drop. A column header or row label the parser cannot identify is not skipped and is not guessed at — it is recorded as an anomaly, and the build refuses to publish while any anomaly is outstanding. The label has to be understood first. This is the single most valuable check in the pipeline: every typo in the table above was found by it, not by anyone reading 291 bulletins. Silence is not evidence of correctness.

Categories and columns are not a fixed set

A reader reasonably assumes the bulletin has five chargeability columns and a stable list of categories. It has neither, and pretending otherwise is how a history quietly becomes wrong. Every column State has ever printed, measured across all 291 bulletins:

A country gets its own column only when demand from applicants chargeable to it exceeds the per-country limit — so columns appear and disappear as demand does. Dominican Republic has one for 8 months and never again. Vietnam has one for 41 months of EB-5 oversubscription. "El Salvador, Guatemala & Honduras" is one column covering three countries. Even India is absent from some 2003 charts, which is why it appears in 284 bulletins rather than all 291. Each is normalized to its own key and simply does not appear in the months State did not publish it. None is dropped for being inconvenient, and none is invented for a month it did not exist.

Category labels drift the same way — 1st, 2A* and 3rd in the early years become F1, F2A and F3 later — and the same bulletin sometimes spells a column one way in its family chart and another way in its employment chart. 1st also means F1 in a family chart and EB-1 in an employment chart, so a label is never resolved without knowing which chart it sits in. 9 categories in this record were published for a while and then retired entirely — including Schedule A Workers and Iraqi & Afghani Translators, which exist for a slice of the mid-2000s and are easy to miss completely. They get no page here, but every cut-off they ever carried is kept.

The one that is genuinely hard: EB-5 in 2022

The EB-5 Reform and Integrity Act of 2022 restructured the fifth employment-based preference. It did not rename a category; it replaced one definition with several. Here is what State actually published, read out of the record rather than summarized:

Two things in that table are worth pausing on. May 2022 is a single-month category — twice. For exactly one bulletin, State printed Unreserved as 2 separate rows with two different cut-off dates, because the regional-centre programme had just been reauthorized and carried its own date. Collapsing them into one category would have silently thrown away one of two real numbers. They are kept apart, and from the following month onward State published a single EB-5 Unreserved row, which is what this site shows now.

And the pre-2022 EB-5 rows are deliberately not joined onto the post-2022 ones. They could be — the chart position is the same, and a longer line looks better. But the fifth preference of December 2001 and the EB-5 Unreserved of today are different categories with different rules, and splicing them would manufacture a 24-year history for a category that does not have one. So the EB-5 pages here start in May 2022 and say why. A short history that is true beats a long one that is not, and on this dataset that judgement is the difference between a chart and a fabrication.

What is a fact here, and what is an estimate

This site answers two different kinds of question and does not let them blur, because only one of them has an honest answer.

"Is my priority date current?" is a fact. It is the comparison of two published numbers — your priority date and the cut-off State printed — and it involves no model, no assumption and no pace. If your date is earlier than the cut-off, it has been reached in that chart. That question is answered for every combination on this site, including ones that are stalled or unavailable, because refusing to answer a question we can answer exactly would be its own kind of dishonesty. The same goes for how far a date is from a cut-off: that is subtraction, and it is always safe to state.

"When will it become current?" is not a fact. Nothing in this data can answer it. Where a cut-off has actually been advancing, this site takes the pace it has been advancing at over the trailing published bulletins, projects that pace forward, and names the bulletin it would reach a given date in. That is an estimate, it is labelled one every single time it appears, and the caveat travels with the number rather than living in a footnote. It assumes a pace holds. Nothing makes a pace hold.

The estimate is also rounded up, always, to a whole bulletin — and never to zero. That sounds pedantic and is not: rounding to the nearest bulletin told every priority date within half a month's pace of the cut-off that it was already current in the current bulletin, in a band up to 54 days wide. That band is precisely where real people are, because people check when they are close.

When this site refuses to estimate, and why

A refusal is always safe. A fabricated date is not. So where the data cannot carry an estimate, this site shows no number at all and states the reason in plain English — it never falls back to a different number, and it never quietly shows a blank. There are five reasons, and each one exists because of a specific way this data misleads a naive model:

In the July 2026 bulletin, State published 150 of the 270 combinations this site tracks. Of those, 90 are advancing at a pace that supports an estimate. The other 60 get no number and a reason. The 120 combinations State did not publish this month get no page either — but every cut-off they ever carried is still in the archive, because a category going quiet is not a reason to erase its history.

Limitations, stated plainly

• This is not legal advice. It is a reference to a published government table. It cannot tell you what will happen to your case, what to file, or when to file it. No honest reading of this data can. For advice about your situation, consult a licensed immigration attorney.
• An estimate is an estimate, not a prediction. Where this site projects, it projects a measured pace forward and assumes it holds. That assumption is not evidence, and the projection is not a schedule, a promise, or a date to plan around.
• Being current in a chart is not the same as a visa being issued. A cut-off passing your priority date means State has reached your place in the queue. It does not mean a visa number is available to you, that your case is approved, or that anything has been issued. The bulletin governs the queue, not the outcome.
• The two charts are not interchangeable, and this site does not decide which one applies to you. Final Action Dates is when a visa can actually be issued; Dates for Filing is when an application may be submitted, and it is usually the more optimistic of the two. Which chart U.S. Citizenship and Immigration Services will accept for adjustment-of-status filings is announced by USCIS each month — not by State, and not here.
• Cut-offs move backward, and this is the single most important thing on this page. When more people apply than the annual limit allows — often because a period of rapid advancement drew filings in — State pulls the cut-off back. It has happened 359 times across the record this site holds. The largest took F3 applicants chargeable to Mexico from 15 October 1993 back to 1 January 1981 in a single bulletin (August 2006) — 12.79 years backwards, undoing more than a decade of queue position overnight. Backward moves are detected and shown as backward moves; they are never smoothed, clamped or averaged away. This is why nothing here is presented as a schedule.
• The archive has 5 holes and a December 2001 floor. Nothing before December 2001 is available under State's URL scheme, and March 2009, September 2009, October 2009, November 2009, October 2012 were never captured by anyone. Neither is reconstructed.
• This site can trail the State Department on bulletin day, for the reasons set out above. Every page names the bulletin month it is reporting; that stamp is the thing to check, and the official bulletin at travel.state.gov is always the authority.

Corrections

Every figure here is traceable to a specific bulletin and stored with the exact cell text State printed, which means an error is findable and fixable rather than arguable. If a cut-off, a movement, a history or a date on this site does not match the bulletin it cites, that is a bug and we want to know: get in touch with the category, the chargeability column and the bulletin month, and it will be checked against the source. Corrections come before everything else here.

The figures on this page were last reviewed against the pipeline on 16 July 2026. The counts themselves are not reviewed — they are recomputed from the data on every build, which is the point.