Puzzle Maker Online

How to Pick a Sudoku Book That's Worth the Money

A good sudoku book is decided by five physical things and one invisible one: difficulty progression, printed grid size, paper weight, binding, answer placement — and whether every puzzle was machine-checked for a single solution. Almost nothing else on the cover matters. "1,000 puzzles!" tells you nothing useful. "Volume 47" tells you less. If you learn to check those six things in the thirty seconds you have while holding a sudoku book in a store, or scrolling the preview images online, you will stop buying books that sit half-finished in a drawer.

This page is a buyer's checklist rather than a review roundup. I'm not going to name titles or publishers, partly because the good ones change constantly and mostly because the same publisher will put out one excellent series and one terrible one in the same year. The criteria hold up better than the recommendations do.

The six things that decide whether a sudoku book is good

Judge a book on progression, grid size, paper, binding, answers, and verification — in that order. Here's the whole checklist in one place, with the failure mode each one guards against.

What to checkWhat good looks likeWhat it prevents
Difficulty progressionClearly labeled tiers, gradual ramp, roughly even counts per tierBuying 200 puzzles you can already do in four minutes
Printed grid size3 inches (75 mm) or larger per gridPencil marks that won't fit in the cells
Paper weight and color70 gsm or heavier, off-white or creamInk bleed-through, glare, erasing that tears
BindingSpiral, wire-o, or a genuinely lay-flat glue bindingFighting the spine with one hand while writing
Answer placementBack of book, indexed by number, never on the facing pageAccidentally reading the solution you were three moves from
Uniqueness checkingStated in the front matter, or verifiable by samplingUnsolvable or double-solution puzzles that make you feel stupid

Everything below expands on those six. If you only remember two, make them progression and grid size — they're the ones people regret most.

Difficulty progression matters more than puzzle count

The single most common flaw in a cheap sudoku puzzle book is that all the puzzles are roughly the same difficulty, regardless of what the section headers claim. You'll see "Easy," "Medium," "Hard," and "Expert" printed at the top of each section, and you'll solve one from each and find them nearly identical.

This happens because difficulty labels are only as good as the generator behind them. A properly rated puzzle is graded by running a solver that knows a fixed list of techniques and recording the hardest technique it actually needed. A lazily rated puzzle is graded by clue count alone, which is a poor proxy — a 30-clue puzzle with all its clues bunched into three boxes can be far harder than a 26-clue puzzle with clues spread evenly.

Here's what a sensible ramp looks like, and what each tier should demand of you:

TierHardest technique neededTypical cluesRealistic time
EasyNaked singles, crosshatching36–454–8 minutes
MediumHidden singles, naked pairs30–358–15 minutes
HardPointing pairs, box/line reduction, triples26–3015–30 minutes
ExpertX-Wing, XY-Wing, coloring22–2625–60 minutes
ExtremeSwordfish, chains, uniqueness arguments20–2445 minutes and up

How to test progression in thirty seconds: open the first puzzle in the easiest section and the first puzzle in the hardest section, and count the givens in each. If the difference is fewer than eight clues, be suspicious. Then look at the hardest section's clue distribution — if every puzzle there looks like it has clues sprinkled evenly across all nine boxes, the book is probably one difficulty wearing four hats.

A book with 200 well-graded puzzles beats a book with 1,000 flat ones every time. You are buying a curve, not a quantity.

Printed grid size, and why 3 inches is the floor

A sudoku grid printed smaller than about 3 inches square cannot hold legible pencil marks, which rules it out for anything above medium difficulty. This is pure geometry. A 3-inch grid gives each cell about 0.33 inches, which is just enough for a 3x3 arrangement of small candidate digits if you write neatly with a sharp pencil. Shrink the grid to 2 inches and each cell is 0.22 inches — you physically cannot fit nine candidates in there.

That's why the six-puzzles-per-page format is a trap. It looks like value. It's fine for easy puzzles you'll solve by scanning, and useless for anything where you need to write candidates, which is everything from hard upward.

Puzzles per pageApproximate grid sizeSuitable for
15–6 inchesAny difficulty, large print, low vision
23.5–4 inchesAny difficulty — the sweet spot
42.5 inchesEasy and medium only
62 inchesEasy only, or travel-size scanning

Large-print editions usually mean one puzzle per page with thick grid lines and a heavy sans-serif numeral. They're marketed at older solvers, but honestly they're the best format for anyone doing expert-level puzzles, because you have room to erase and rewrite a cell four times without shredding the paper.

Check the cover claim against reality: some books say "large print" and only enlarge the page, not the grid. Look at the preview image and compare the grid to the page edge.

Paper, ink, and the erasing problem

Sudoku paper needs to survive repeated erasing, and most cheap puzzle books use paper that doesn't. Two numbers matter: weight and opacity.

The practical test if you can hold the book: open to a middle page and hold it up to a light. If you can clearly read the grid on the other side, the paper is too thin for pencil work.

If you write in ink — and plenty of confident solvers do — none of this matters as much, but bleed-through still does.

Binding decides whether you'll actually use it

A sudoku book that won't stay open is one you'll abandon. You need one hand for the pencil, which leaves one hand for the book, which means the binding has to hold the page flat on its own.

If a book is perfect-bound and you love it anyway, cracking the spine along the center works, though you'll eventually shed pages.

Where the answers should be

Answers belong at the back of the book, numbered to match the puzzles, and never on the same spread as the puzzle itself. This sounds trivial until you've had your eye land on a solution grid while turning a page and lost twenty minutes of work in one glance.

Three layouts you'll see, ranked:

  1. All solutions in a numbered answer section at the back. Correct. Ideally printed small, in a light gray, so a casual flip past doesn't burn anything into your memory.
  2. Answers upside down at the bottom of the page. Tolerable in newspapers, bad in books — peripheral vision picks up more than you think.
  3. Answers on the facing page. Genuinely disqualifying. Skip the book.

A related detail that separates careful publishers from careless ones: whether the answer key is checkable. Some books print only the completed grid, which is all you need. A few print nothing at all for the hardest tier, which is inexcusable.

The invisible criterion: unique, machine-checked solutions

Every legitimate sudoku puzzle has exactly one solution, and a puzzle with two solutions or none is a manufacturing defect that no amount of skill will let you solve. This is the failure you cannot see from the cover, and it's the reason some books generate a stream of one-star reviews saying "puzzle 143 is impossible."

A puzzle ends up broken in one of two ways. Either a clue was mistyped in layout, producing a grid with no solution at all, or the generator removed clues without re-testing uniqueness, leaving a grid that accepts two valid completions. The second kind is nastier, because you can solve most of it and then hit a point where two digits could go either way and pure logic can't choose.

The pattern to recognize: four cells sitting at the corners of a rectangle, spanning two rows, two columns, and exactly two boxes, holding only two candidate digits between them. If you reach that and nothing else in the grid constrains it, the puzzle has at least two answers. That configuration is called an unavoidable set, and a generator that tests properly can never produce one unresolved.

Good publishers say something in the front matter like "all puzzles computer-verified for a unique solution." Take that as a positive signal. If a book says nothing, sample two or three of the hardest puzzles before committing to the series.

For what it's worth, this is one place where digital beats print outright — a generator can verify uniqueness on every single puzzle it produces, which is exactly what the free maker at puzzlemakeronline.com does before it prints anything.

Matching the sudoku book to the person

The right book depends far more on who's solving than on how many puzzles are inside.

Printing your own instead

If you want a specific difficulty, a specific grid size, and a guaranteed unique solution on every puzzle, printing is cheaper and more controllable than buying. You choose the tier, you choose how many grids per page, you print on paper you already like, and you can run off ten more when you finish.

The obvious trade-off is that a printed stack isn't a book. No binding, no cover, nothing to hand someone as a gift. But for a solver who has landed on one difficulty and one format and wants a hundred more of exactly that, it's hard to beat. Our free puzzle maker generates sudoku at several difficulty levels with an answer key on every sheet, and if you'd rather construct puzzles from scratch, make your own sudoku walks through the whole process.

Frequently asked questions

How many puzzles should a sudoku book have?

Between 100 and 300 is the practical range for a single volume. Below 100 you'll finish it too fast to justify the shelf space; above 300 the book gets thick enough that perfect binding stops lying flat and the grids shrink to fit the page count. Puzzle count is the least important spec on the cover — a 120-puzzle book with a real difficulty curve is worth more than a 1,000-puzzle brick of identical mediums.

Are the puzzles in cheap volumes actually different from expensive ones?

Usually not in origin — most publishers license puzzles from the same handful of generation engines. The difference is in curation and production: whether someone graded the puzzles honestly, whether the grids are big enough to write in, and whether the paper survives an eraser. You're paying for editing and manufacturing, not for rarer puzzles.

What does "computer-verified" mean on a sudoku book cover?

It means each puzzle was run through a solver to confirm it has exactly one solution before printing. That's the minimum bar for a legitimate puzzle and it's worth looking for, because a puzzle with two valid completions is unsolvable by logic no matter how good you are. It says nothing about difficulty accuracy, which is a separate check.

Should a beginner buy a mixed-difficulty book or an all-easy one?

Mixed, as long as the ramp is real. An all-easy book gets boring in a week and teaches you only scanning. A book that runs easy through hard gives you somewhere to go, and the moment you first need a technique beyond scanning is the moment sudoku becomes interesting. Our sudoku rules and tips and tricks pages cover the transition.

Is a sudoku puzzle book better than an app?

They're good at different things. A book has no timer, no hints, no notifications, and no screen at bedtime, and the physical act of writing seems to help a lot of people concentrate. An app auto-fills candidates, checks your work, and never runs out. Most regular solvers end up using both — an app for idle minutes, a book for a deliberate half hour.

What grid size should I look for in a large-print edition?

At least 4 inches (100 mm) square, and preferably 5 to 6 inches with one puzzle per page. Also check the numeral weight: a thin serif face at 18 points is harder to read than a bold sans at 14. And make sure the grid is enlarged rather than just the page trim size, which is a common bit of cover trickery.