Wordle Answer Archive: Every Past Word by Date and Puzzle Number

How Wordle Chooses the Answer Each Day

Wordle does not invent a word each night, and it does not pick at random from a live dictionary. It walks through a fixed, edited list in order. Josh Wardle built the original list by filtering a large dictionary down to common five-letter words and then removing anything he judged obscure, rude, or hard to justify on a puzzle a family might open at breakfast. When the New York Times acquired the game in early 2022 it kept that list and the existing numbering, and its editors have adjusted it since — dropping a few words here, adding a few there — so the rotation keeps running without a gap.

The edits are the interesting part, because they explain most of the surprises players complain about. Regular plurals are excluded, which in practice means four-letter words with an S bolted on: CATS, LAKES, BANDS will never be an answer. Proper nouns are out. Abbreviations and initialisms are out. Slurs, crude body parts and anything a school would object to were removed at the very start and have not come back.

What remains is a pool of everyday English that a general audience can recognise on reveal, even if they would never have guessed it cold. That is the whole editorial standard: recognisable once revealed, not obvious beforehand. It is a deliberately low bar, and the gap between "low bar" and "easy" is the entire experience of a hard puzzle. The solver and the hint techniques both exist to help you play well inside that gap.

What "the Answer List" Actually Means

Almost every confusion about Wordle answers traces back to one distinction: the game runs two separate lists, and they are not the same size. The first is the answer pool — the words that can be chosen as the daily answer. The second is the acceptance dictionary, a much larger list of every legal guess, which exists so that a near-miss like ADIEU or SOARE is accepted as "a word, but not today's" instead of being rejected as a typo.

Both lists take five letters, and that shared shape is what fools people into treating them as one list. The consequences are immediate and practical. A tool that filters the acceptance dictionary will cheerfully return a dozen words that fit your colours and can never be the answer. A tool that filters the answer pool returns fewer, stranger-feeling words, and those are the real candidates. The distinction is also why the answer list here is small enough to print in full while the guess dictionary runs to well over ten thousand entries.

So when you read a claim like "there are 2,300 Wordle answers", understand that it refers to the pool, not to everything you are allowed to type. The gap between those two numbers is the single most useful fact in Wordle strategy, and it is the reason a solver should switch to answer-pool filtering the moment you are close to the end of a board.

How Puzzle Numbers Map to Dates

The number in the game's header is a day counter, not a random identifier. Wordle's first published puzzle is dated 19 June 2021, and every subsequent calendar day adds exactly one. That makes the mapping pure arithmetic: subtract the origin date, divide by the number of milliseconds in a day, and you have the puzzle number. This archive computes it the same way, which is why you can check any row against your own screen.

Two caveats are worth understanding. First, the counter is anchored to a date, not to a time of day, so daylight saving does not disturb the arithmetic — but it does shift the gap between the UTC clock and your local midnight by an hour twice a year. Second, and more practically, the game flips at midnight on the device you are using. A player in Sydney can be a full day ahead of a player in Los Angeles, and both are correctly on "today's" puzzle. Spoiler posts usually reference US Eastern because the Times is in New York, which means a word shared during a New York afternoon is often tomorrow's word for readers in Asia.

The anchor points below use the same origin as the calculation on this page, so you can verify any row against a date you remember:

The Rules the Answer List Follows

Three editorial rules do most of the work, and you can see all three directly in the table above. The first is that regular plurals are excluded. It is easy to over-generalise this into "no word ends in S", which is false — the list contains FOCUS, GLASS, CHAOS, ATLAS, BASIS, CRASS and two dozen more. What it excludes is the suffix, not the letter. If a word merely happens to end in S because S is part of its shape, it is perfectly eligible.

The second rule is that no answer repeats until the pool has been exhausted. This was true for an extraordinary run: every one of the first 1,689 puzzles used a word that had never been used before. Then the original pool came to an end, and the rotation began cycling. Puzzle 1,689 was played on 1 February 2026, and puzzle 1,690 — the very next day — was CIGAR again, the same word that opened the game in June 2021. Since then the archive has picked up a number of repeat appearances, all of them old answers returning rather than new words entering.

This is worth internalising because it changes how you read a familiar word. A player who meets a word they recognise is not necessarily looking at a wrong spoiler — they may genuinely be on a recycled puzzle. The check is always the puzzle number: if the number matches your screen, the word is right even if you have seen it before. The third rule is that the guess pool is far wider than the answer pool, which we covered above and which explains why so many "reasonable" words turn out to be impossible.

What the Archive Statistics Tell You

The numbers above are computed from this list as you load the page, so they stay true as the file grows. The total is the number of recorded answers. The span is the distance in days between the first and most recent entry. The unique count is how many genuinely different words those answers represent — and the gap between that figure and the total is the recycling, quantified rather than asserted.

The first-letter spread is more interesting than it sounds. Across the whole list only a limited set of letters ever open an answer, and the letters that have never opened one are —. If you open with a word starting one of those letters, you are testing a letter the rotation has never once used in that position across every recorded puzzle. Knowing which letters are common in first position is worth more than any frequency table you can memorise from a generic word list, because the answer pool is a curated subset and its first-letter distribution is nothing like the dictionary's.

The duplicate-letter and vowel counts round out the picture. A meaningful minority of answers contain a doubled letter, which is why duplicate detection is not an edge case but a routine part of solving. A smaller but still substantial share carry three or more vowels, which is the family you should suspect whenever your board feels vowel-starved and every obvious vowel has come back grey. Both figures argue against the common strategy of assuming a light, simple word: the pool is more varied than intuition suggests.

How to Use This Archive to Build Your Own Answer Memory

The most valuable use of an archive is not answering "what was the word on Tuesday" six months later. It is training your intuition so that next Tuesday's puzzle is easier. The mechanism is simple: browse by first letter rather than by date, and you are effectively seeing the shape of the answer pool — which letters open words, which letters appear in the middle, which endings cluster. Ten minutes of that reshapes how you read a fresh board.

A second technique is to filter by year and compare. Look at a single year's worth of answers in one sitting and you will notice recurring families you had not registered: the -ATCH cluster, the -IGHT cluster, words with a doubled middle consonant. Those families are not random; they exist because the pool was built from a curated English vocabulary, and vocabulary comes in families. A player who can name the family their board implies converts a wall of greens into a two-word decision.

The third technique is the one most people skip and the one that pays off fastest: use the archive to check your own recall. Cover a past answer, write down what you remember, then look it up. The gap between recall and reality is a direct measurement of your board-reading skill, and it is far more informative than your win rate because it isolates reading from luck. Do twenty of those and you will find that your errors are not random — they concentrate in duplicate-letter boards and in vowel-light boards — which are precisely the two shapes you can then practise deliberately in the practice game.

Why Tracking Your Own History Helps

Every Wordle client keeps a statistics panel, and most players read it as a scoreboard: games played, win percentage, current streak, longest streak, and a guess distribution. Read it as a diagnostic instead and it becomes much more useful. The guess distribution in particular tells you where your turns actually go. A player who wins mostly in three but occasionally needs six has a ceiling problem, not a floor problem — they are fast on easy boards and stranded on hard ones, and the fix is elimination discipline rather than opener choice.

Keeping your own log adds the dimension the built-in panel throws away: what the word was. A note with the date, the puzzle number and your finished grid is more durable than browser storage, which a cache clear will wipe, and it is the only record that lets you do the recall drill above. It also settles arguments. When two people remember the same puzzle differently, the grid is falsifiable and memory is not — the colours decide, not the recollection.

There is a third reason that is less obvious. Reviewing your own history exposes the specific puzzle shapes you handle badly, and those shapes are not evenly distributed. If you lost streaks on boards with four greens and four viable candidates, you have an endgame problem and the -ATCH family is where to practise. If you lost them on boards where you had four greens by guess two and then wasted turn three confirming rather than eliminating, you have a turn-allocation problem. Either way the fix is specific, and you cannot find it without a record.

How to Verify a Claimed Past Answer

Before you trust any word — from this archive, from a group chat, or from a spoiler thread — check it against colours you recorded at the time. This takes about twenty seconds and settles the question without relying on anyone's memory, including your own. Enter your greens, yellows and greys into the Wordle Solver and see whether the claimed word survives. If it does not, the claim belonged to a different puzzle number, a different game, or nobody's board at all.

Two specific checks catch most mistakes. First, confirm the puzzle number: a spoiler one day off is by far the most common complaint, and for readers outside US Eastern it is the default rather than the exception. Second, count the letters. If someone claims a word with two Rs and you remember a green R beside a grey R, the claim is wrong regardless of how authoritative it sounds — a grey copy next to a coloured copy means the answer contains exactly one.

It is also worth knowing what this archive is not. It is a record of published answers, not a record of guesses, and it cannot tell you whether you solved a given puzzle or how many tries it took you. It covers every answer up to the most recent update and grows daily; it is not a guarantee about any future puzzle, and it is not a substitute for the official archive if you need to replay a specific board.

How to Read the Table Above Without Spoiling Yourself

Every answer in the table starts blurred, and the blur is not decoration. The text is genuinely hidden from screen readers until you reveal it, so a screen-reader user who is simply browsing the table is not handed the puzzle by accident. That matters more than it sounds, because the most common way people meet this kind of page is by searching for a date they half remember — and being handed a list of every answer in order while you try to find one row is a spoiler in its own right.

There are three sane ways to use the table. The first is deliberate lookup: you search for a date, reveal the single row you came for, and move on. The second is study: hit Reveal all and read a single year's worth, preferably filtered to one first letter, purely to build a feel for the pool. The third is recall testing, where you keep the words hidden, write down what you remember, and then reveal to check yourself. The third is the one that actually makes you better at the game, and it is the only one that leaves any learning behind.

What you should not do is scroll the unfiltered, revealed table and then open the game. That produces a finished board and no improvement, which is a worse outcome than either playing or studying honestly. If your goal is a streak, the table is a legitimate rescue on a day you have already lost. If your goal is a better score, the ladder on the hints page is the better stopping point, because it stops one click earlier than the answer.

Honest Limits of This List

This archive covers every published answer from 19 June 2021 up to the last update, and it is regenerated automatically rather than maintained by hand. That gives it two properties worth stating plainly. It will not contain a fabricated or predicted word, because every entry is read from the official source at the time the puzzle resolves. And it will not silently drift, because a page that prints a puzzle number is making a checkable claim — if an entry here ever disagreed with the word in your own game, that would be a bug worth reporting immediately.

It also has limits that no amount of care removes. The list is only as complete as its most recent update, so on the day a new puzzle appears there may be a short window before it is added. Historically, entries come from observed public results rather than from an official feed of record, which means an error would be invisible unless someone checks it against a board. Finally, the answer pool itself is edited over time by the publisher, so this is a record of a moving target rather than a fixed historical object — a word removed from the pool stays in this archive, because the puzzle that used it really did use it.

How the archive is built, and how you can tell

Nothing in the table above is typed in by hand. The archive is a data file that an automated job refreshes each day, keyed by the US Eastern calendar date — the same keying the game itself uses — and the page renders it from that file. That is worth understanding, because it explains why every row can carry a puzzle number without anyone doing the arithmetic twice.

The puzzle number is derived, not stored. It is the number of days since 19 June 2021, plus one, which means the date column and the puzzle-number column cannot drift apart. If a date were missing, duplicated or out of order in the source, the numbering would visibly skip or repeat in the table, and the gap in the puzzle column is the giveaway. You can check any row against a date you remember: puzzle 1 is 19 June 2021, puzzle 1,000 fell on 15 March 2024, and puzzle 1,948 fell on 18 October 2026.

One consequence of that setup catches people out the first time they look at the table. The file carries the pre-published answers for the days ahead as well as the ones already played, so the date range shown under the heading can end on a date that has not happened yet. Those rows are real announced answers, not predictions, and their puzzle numbers are already correct — but they are future puzzles, so treat them as the archive's forward edge rather than as a record of something you can go back and verify against a finished board.

The statistics are computed the same way, from the same rows you are looking at, in your browser as the page loads. The total, the span in days, the count of distinct words, the number of letters that have ever opened an answer, the repeated-letter count and the three-or-more-vowels count are all derived from the list rather than typed into the page. That is why the strip above the table cannot disagree with the table itself, and it is why the numbers move on their own as the file grows.

Looking up a date: three ways that work

The search box accepts more than a word, and knowing the three forms it recognises saves a lot of guessing. Type an ISO date such as 2023-05-11 and the table narrows to that exact row. Type a year such as 2023 and it narrows to that year's puzzles. Type a four-digit puzzle number and it finds that row, which is the most useful option when the number on your own screen is the thing you remember. The year dropdown gives a count beside each year so you can see the size of each season before you commit to it.

The row of letters under the controls filters by the first letter of the answer. Letters that have never opened an answer in the whole recorded list are dimmed, and the count for each letter is in the button's tooltip, so the row doubles as a quick map of how the answer pool is distributed. The Reset button clears the search box, the year filter and the letter filter in one press, and it also re-hides any words you had revealed.

Answers stay blurred until you ask for them, and that applies to the table below as much as to the rows above: the text is genuinely hidden from screen readers until it is revealed, so a screen-reader user browsing the table is not handed a puzzle by accident. Click any single word to reveal just that one, or use Reveal all when you are deliberately reading the list rather than trying to solve.

The 2026 recycling, measured rather than asserted

The single most striking fact in this archive is that it stopped being a record of entirely new words. For the first 1,689 puzzles, every answer was a word that had never been used before. Then puzzle 1,690, played on 2 February 2026, was CIGAR again — the same word that opened the game on 19 June 2021.

It is tempting to picture what follows as a closed loop, but the list is more interesting than that. Of the 305 answers listed from puzzle 1,690 onwards, 273 are words that had never been played before and 32 are second appearances of words already in the archive. The pool was topped up rather than merely looped, and the recycling runs alongside new material instead of replacing it. Each of the 32 recycled words has now appeared exactly twice, and no answer has yet been played a third time.

There is a pattern in which words come round again: every one of the 32 recycled answers first appeared early in the game's history, within the first 1,227 puzzles. They are the familiar ones — CIGAR from puzzle 1, then SQUAD, AWAKE, LINEN, GRADE, IVORY, COMET, STAND, BRING, CLOCK, AGREE, SMILE, THUMB, BREAK, ACUTE, BATON, STOUT, FLUME, TRACE, CLICK, FOCUS, FORTH, PROVE, MAJOR, STINK, MOTOR, FIRST, BLURT and SPICY among them. The recent stretch of the archive is being drawn from the early stretch, which is why a word you have definitely played before can turn up on a puzzle that feels new.

Across the 1,994 answers recorded so far, 1,962 are distinct words. The gap of 32 is the recycling, counted exactly. This changes how you should read a familiar word: a player who meets one they recognise is not necessarily looking at a bad spoiler, because they may genuinely be on a recycled puzzle. The check is always the puzzle number — if the number on your screen matches the number in the table, the word is right even though you have seen it before. For preparation, the same fact is useful in reverse: the words most likely to come round again are the well-known early ones, and studying the first few hundred puzzles tells you what the recycling draws on.

When the archive and your screen disagree

Occasionally a claimed answer and your own board will not agree, and the archive can settle it in about twenty seconds. Enter your greens, yellows and greys into the Wordle solver and see whether the claimed word survives the colours you recorded. If it does not, the claim belonged to a different puzzle number, a different game, or nobody's board at all.

Two specific checks catch most of the mistakes. First, confirm the puzzle number, because a spoiler that is one day out is the most common complaint, and for readers outside US Eastern it is the default rather than the exception. Second, count the letters: if someone claims a word with two Rs and you remember a green R beside a grey R, the claim is wrong however authoritative it sounds, because a grey copy next to a coloured one means the answer holds exactly one.

And if you are checking today's board rather than a past one, stop and read the daily answer page and the hint ladder first. The ladder stops one click earlier than the answer, and a hint that leaves you something to do is worth more than a word that ends the puzzle. The full archive is the right tool for a date in the past and the wrong tool for a date you have not finished.