The State of Word Games 2026: 12,478 Five-Letter Words, Measured

By Ramkrishna Prasad · 4 November 2026

This report is about the raw material of word games, not the people who play them. Every number below comes from one file: our own list of 12,478 five-letter words, the pool behind the solvers and five-letter pages on this site. We counted letters, positions, vowels, repeats, three-letter openings and endings, and alphabet weights. The result is a reference page that teachers, researchers, and journalists can check and cite.

The title borrows the language of industry reports, so scope matters: this is not a survey of the market, player counts, or revenue. It is a census of the words. Where the data supports a practical observation, we say so; where it does not, we say that too.

Executive summary

• The list contains 12,478 five-letter words and 12,478 distinct words, with no duplicates, giving 62,390 letter slots to count.
• E is the most common letter overall with 6,441 appearances (10.32%), only 14 ahead of S at 6,427 (10.30%). The top five letters account for 43.09% of all letters.
• 59.42% of words contain exactly two vowels (A, E, I, O, U), 29.94% contain one, and 51 words (0.41%) contain none.
• 4,465 words (35.78%) repeat at least one letter, while 1,727 (13.84%) contain a consecutive double letter.
• S ends 3,805 words (30.49%) and starts 1,521 (12.19%), making it the most common letter at both edges of the list.

Methodology

The source is wordlist-5.js, AnyWordSolver's five-letter acceptance list: 12,478 entries, each exactly five letters and using only A–Z. The file needed no cleaning and contained no duplicates. We loaded it in Node.js with a minimal window shim, uppercased every entry, and counted the full set on 4 November 2026.

Definitions change percentages, so they are stated plainly. A vowel is A, E, I, O, or U; Y is a consonant throughout. A repeated letter appears at least twice, adjacent or not. A consecutive double needs the same letter in neighboring positions, as in ABBEY. A three-letter beginning is the first three characters; an ending is the last three. Letter weight assigns A=1 through Z=26 and adds the five values.

Counts are exact integers; percentages are rounded to two decimals, so columns may not sum to exactly 100%. Positional ties are alphabetical. The list is a solver pool rather than a corpus of everyday speech: it includes rare, archaic, and borrowed words. That is the pool a five-letter game must work with, but it limits what the numbers say about ordinary English.

The list at a glance

A fixed-length list makes the denominators simple: 12,478 words produce 62,390 letter slots, and every percentage here is a share of one of those totals. There is nothing to weight for length and no missing data.

The last row is the striking one. Nearly two-thirds of the list — 8,013 words — uses five different letters. A guess with no repeats therefore matches the majority of the pool rather than compromising. The remaining 4,465 words are where repeats live, and we return to them below. The full five-letter list and the no-repeated-letters subset are both browsable.

Letter frequency: E and S finish almost tied

The list is not close to uniform. If every letter appeared equally often, each would take 3.85% of the 62,390 slots and appear about 2,400 times. E and S each appear more than 6,400 times, while Q appears 104. The top five letters — E, S, A, O, R — take 43.09% of all letters between them.

The near-tie at the top is the kind of detail a single decimal place would hide. E leads S by 14 appearances out of 62,390, or 0.02 percentage points. Calling E more common is technically true for this list, but the honest reading is that the two are level. A at 5,738 sits clearly behind them, and the drop continues through O and R.

The bottom is more lopsided. Q, X, J, Z, and V together account for just 2.75% of all letters. Q's 104 appearances mean fewer than one word in a hundred contains one. That is why an opening built around E, S, A, O, and R covers far more of the average word than one built around high-value Scrabble tiles; the word solver applies the idea to a specific board.

Where letters sit: the positional map

Overall frequency hides a stronger pattern: the same letter is not equally likely in every slot. Position 2 is the clearest case. Its five most common letters are all vowels, and together they fill 8,209 words (65.79%). Position 4 is led by E at 2,257 words (18.09%), nearly one word in five.

Read left to right and a shape appears. The first slot is consonant-heavy, the second a vowel zone, the fourth the E slot, and the fifth the S slot by a wide margin: 3,805 words end in S, more than twice the 1,477 that end in E.

That last figure is the largest positional effect in the data. S is 30.49% of final letters; E is second at 1,477 (11.84%) and Y third at 1,254 (10.05%). A five-letter word is unusually likely to end in S, often as a plural or third-person ending.

For guessing, combine the two views. STARE-style openings place common letters where they are common, which is why they recur in strategy discussions. This dataset cannot prove any opening is optimal, because optimality depends on a game's rules and scoring. The word finder turns a partial pattern into candidates.

Starting letters: S owns the first slot

All 26 letters start at least one word, so the number that never starts a word is zero. The distribution is still far from even. S starts 1,521 words (12.19%), more than three times the 3.85% a uniform letter would get. C, B, P, and T follow at 888, 871, 821, and 787. At the other end, X starts 15 words (0.12%), Q starts 76 (0.61%), and Z starts 98 (0.79%).

The first-letter row is also a reminder that "common letter" and "common first letter" are different questions. E is the most common letter overall but starts only 288 words (2.31%) — fewer than F, G, or W. C is only the 13th most common letter overall yet starts 888 words, the second-highest total. A letter's overall count and its positional behavior have to be kept separate.

If you know the first letter, the starting-letter index splits the list 26 ways, and the S page is the largest slice. If you are choosing an opening, the table argues for at least one high-frequency starter and against a guess built entirely from rare initials.

Vowels: two is the norm, none is rare

Using the strict A, E, I, O, U definition, the list is overwhelmingly a two-vowel list. Exactly 7,415 words (59.42%) contain two vowels, and 3,736 (29.94%) contain one. Add the three-vowel group at 1,249 (10.01%) and 69.43% of the list sits at two or three. The extremes are thin: 27 words have four vowels, and none has all five.

The zero-vowel group is small: 51 words, about one in 245. Because Y counts as a consonant, that group includes words whose only vowel-like letter is Y. Counting Y as a vowel would shift the distribution; this report does not, and the definition is stated so the numbers can be reproduced. The four-vowel group is smaller still at 27, and the five-vowel group is empty.

For play, a guess with two vowels matches the largest group by a wide margin. Three vowels is still reasonable, since that group is more than twenty times larger than the zero-vowel group. Four vowels is chasing a rarity. The positional table already showed which slots those vowels favor. The no-vowel word list is the small, strange corner of the pool.

Repeated letters: about a third of the list

Repeats are common enough that treating every word as an isogram would be a mistake. Of the 12,478 words, 4,465 (35.78%) contain at least one letter twice or more. The other 8,013 (64.22%) use five distinct letters. Consecutive doubles are much rarer: 1,727 words (13.84%) contain two identical letters side by side.

The gap between 35.78% and 13.84% is the interesting part. Most repeated-letter words repeat a letter at a distance rather than next to itself. ABACA is the clean example: A appears three times, never twice in a row. ABBEY is the opposite, with a double B in the middle. The list contains no word with three identical letters in a row, so the maximum is three copies, always separated.

This is where a solver's assumptions can go wrong. Assume five distinct letters and you miss roughly one word in three. Assume repeats are always doubles and you miss most of the repeat group, since only 1,727 of the 4,465 repeated-letter words contain a consecutive pair. Keep both possibilities open until feedback forces a choice. The double-letter list and the all-distinct list make the difference concrete.

Three-letter openings and endings

Three-letter chunks are where the list starts to look like a language. Endings are dominated by S: LLS, RES, CKS, LES, NES, and TES are plural or third-person patterns, and STS and NGS extend the family. Beginnings are more varied: STO, CHA, CHI, SHA, and SPA lead, and no opening covers even half a percent of the list.

The most important number in these tables is not in them. The list contains 2,696 distinct beginnings and 2,332 distinct endings. A top-15 list that peaks at 0.51% is a list of small effects, not a shortcut. Even the strongest ending, LLS, accounts for only 64 words. Learning the top endings helps recognition but will not carry a solve on its own.

What the tables show is asymmetry. Endings are more concentrated and grammatical; beginnings are more lexical, spread across ST, CH, SH, SP, and BR. If you have the last three letters, the search space is narrower than with the first three. That describes this list, not a universal rule, and the answer archive makes it visible across many days.

Letter weight: the alphabet score

Assigning A=1 through Z=26 and summing the five letters gives a rough measure of how expensive a word looks. The average is 60.30, the median 60, and the middle half runs from 50 to 71. Five uniformly random letters would average 67.5, so the list sits below that benchmark, as expected from a pool dominated by E, S, and A.

The extremes are a sanity check. ABACA is lightest at 8 because it uses A, B, and C; YUZUS is heaviest at 112 because it carries Y, Z, and two Us. Neither is a proxy for difficulty: ABACA is obscure despite its low weight, and a high weight makes a word unusual, not harder to guess. The metric measures letter values, not frequency, meaning, or skill.

The distribution is tight around the middle: half the list falls between 50 and 71, and the mean and median almost coincide. That is a small argument for treating letter weight as a descriptive curiosity rather than a ranking tool. It is included because it is easy to reproduce from the same list. For a tool that works on letters rather than scores, the word games hub is the place to start.

How to cite this report

If you use these figures in an article, a lesson, or a paper, cite the page rather than an individual table. The suggested citation is:

Prasad, Ramkrishna. "The State of Word Games 2026: 12,478 Five-Letter Words, Measured." AnyWordSolver, 4 November 2026, https://anywordsolver.com/state-of-word-games-2026.html.

Canonical URL: https://anywordsolver.com/state-of-word-games-2026.html. The underlying list is wordlist-5.js, 12,478 entries. Figures were computed on 4 November 2026 and rounded to two decimals. The definitions above are enough to recompute any count not in the tables.

Limitations

• The list is a solver acceptance pool, not a balanced corpus of everyday English. Rare, archaic, and borrowed words count equally, so these are list frequencies, not usage frequencies.
• There is no information about how often a word appears in speech or writing. A word that appears once in the list is treated the same as one a reader meets daily.
• There is no player data: no solve rates, no guess distributions, no timing, and no comparison with any other site or game.
• Y is counted as a consonant. Counting it as a vowel would change the vowel distribution, especially the 51-word zero-vowel group.
• Positional counts are marginal. They describe each slot separately and do not model dependencies between letters, so they cannot give the probability of a full pattern.
• Percentages are rounded, the list can change when the dictionary is updated, and a different list would produce different numbers even with the same definitions.

Related reading: the best daily word games of 2026 · how to get better at word games · the full five-letter word list · the Wordle-style solver.