Jigsaw Sudoku: How Irregular-Shaped Regions Change the Puzzle

Jigsaw sudoku keeps every rule of standard sudoku and changes exactly one thing: the nine boxes are no longer 3x3 squares, they're irregular interlocking shapes. Each region still holds nine cells and still needs the digits 1 through 9 exactly once.

That single change makes the puzzle noticeably harder, and it unlocks one deduction technique that regular sudoku never gives you. It goes by several names — jigsaw sudoku, squiggly sudoku, nonomino sudoku, geometry sudoku — but they all describe the same variant.

The Rules, Precisely

• Every row contains 1-9 exactly once
• Every column contains 1-9 exactly once
• Every irregular region (nine cells, any connected shape, marked by heavy borders or shading) contains 1-9 exactly once

That's it. The row and column rules are untouched. The only difference is that "box" now means an arbitrary connected shape rather than a neat square.

Puzzle designers mark the regions either with heavy borders or with alternating background shading. Shading is considerably easier to read and, if you have a choice, worth preferring — a substantial share of jigsaw mistakes come from misreading which region a cell belongs to.

Why Irregular Regions Are Harder

You lose positional intuition. In standard sudoku, experienced solvers stop consciously reading the box borders. After enough puzzles, the 3x3 grouping is baked into how you see the grid — you know without thinking that cell R4C5 sits in the center box.

Jigsaw regions have to be read fresh on every single puzzle, and every cell requires a conscious lookup. That constant checking is where most of the added difficulty comes from. It isn't harder logic; it's the same logic with a permanent tax on every step.

Regions can span far more rows and columns. A 3x3 box always touches exactly three rows and three columns. A jigsaw region might snake across five rows and four columns. That spread changes which constraints interact, and it means the familiar "this box only touches these three rows" shortcut simply doesn't exist.

A region shaped like a long horizontal strip behaves almost like a row and constrains very little beyond it. A region packed into a compact blob overlaps heavily with two or three rows and columns and constrains a great deal. Reading which kind you're looking at is a real skill in this variant.

Scanning is slower. The number-by-number scan that opens every standard solve still works, but you can no longer sweep three neat rows at a time. You have to trace each region's actual outline, which is slower and more error-prone until you've internalised the shapes on that particular grid.

The Technique Jigsaw Unlocks: The Law of Leftovers

This is the real reward for the added difficulty, and it doesn't exist in standard sudoku.

Take any set of rows that a group of regions almost fills. Suppose the top two rows of the grid are exactly covered by two regions — except that one region pokes one cell down into row three, and one cell of row two belongs to a third region instead.

Rows 1 and 2 together must contain each digit exactly twice, once per row. The two regions together must also contain each digit exactly twice. If the two sets of cells were identical, they'd hold identical contents. They aren't quite identical — they differ by the cell that pokes out and the cell that doesn't belong.

Therefore those two leftover cells must contain the same digit. The cell poking down into row three and the cell of row two belonging to the outside region hold the same value.

In practice this often pins down a single cell outright, and it's frequently the deduction that cracks an otherwise stuck jigsaw grid. Solvers who don't know it find jigsaw puzzles unreasonably hard; solvers who do find them a fair trade.

The same logic works with columns, and with larger sets — three rows against three regions, four against four. The bigger the set, the rarer the clean alignment, but the two-row version comes up constantly.

How to Approach a Jigsaw Puzzle

Trace every region border before you start. Spend thirty seconds following each shape with your finger or a pencil so you know the nine regions as shapes, not as a vague pattern. This pays for itself immediately and prevents the single most common error in the variant.

Look for the most compact regions first. A region packed into a small area behaves almost like a standard box, and its cells share rows and columns heavily — which means strong constraints and early placements. Long snaking regions constrain less per cell and are usually the last to resolve.

Watch for regions that fill a row or column almost entirely. These are where the law of leftovers applies, and they're the highest-value places to look when progress stalls. Scan for them deliberately at the start so you know where they are before you need them.

Pencil mark early. Everything from standard strategy still applies — naked singles, hidden singles, pairs, pointing sets — and all of it depends on accurate candidates. Because region membership is harder to eyeball, jigsaw rewards marking sooner than a standard grid of the same rated difficulty would.

Mark the region boundaries on your own copy if the printed shading is faint. Two minutes with a highlighter is cheaper than one misattributed cell.

Standard vs. Jigsaw at a Glance

How the Standard Techniques Translate

Cross-hatching works but is slower, because striking out rows and columns against an irregular region means tracing its outline rather than glancing at a square.

Naked and hidden singles work identically. No adjustment needed beyond correct region attribution.

Naked pairs and triples work identically within any unit — row, column or region.

Pointing pairs work, and are genuinely harder to see. In a standard box, a candidate confined to one row is visually obvious because the box is three cells wide. In a jigsaw region spanning five rows, spotting that a candidate appears only in cells sharing a single row takes deliberate checking.

X-Wing and swordfish work unchanged, since they operate on rows and columns rather than regions.

Common Mistakes

Reading a cell into the wrong region. By far the most common error, and it poisons every deduction that follows. When regions have similar coloring or thin borders, double-check membership before eliminating a candidate.

Assuming a 3x3 area is a region. Some jigsaw puzzles include a region that happens to be square or nearly so. It's easy to then treat neighboring cells as though the whole grid were standard. Trust the printed borders, not the shape you expect.

Ignoring the leftovers technique. Most solvers who bounce off jigsaw sudoku have simply never learned it and conclude the variant is unfairly hard. It isn't — it just has its own tool, and without it some puzzles genuinely won't yield.

Applying box-shaped intuition to a snake. A region spanning six rows shares almost nothing with any single row, so the mental shortcut "this region will resolve when these three rows fill" is simply wrong.

Where Jigsaw Sits in Difficulty

A jigsaw puzzle with the same number of starting clues as a standard puzzle will generally play harder, so well-made jigsaw sets often include a few extra givens to compensate.

If you're comfortable with hard standard sudoku, start on medium jigsaw rather than matching your usual level. The adjustment is real, and it's mostly about retraining how you read the grid rather than learning new logic — which means it comes quickly, usually within five or six puzzles.

A sudoku puzzle maker that supports variant grids lets you print jigsaw puzzles at a chosen difficulty, which is a much gentler way in than picking up an expert-rated one cold from a puzzle book.

Other Variants Worth Knowing

Jigsaw sits in a family of region-based variants:

Diagonal (X) sudoku keeps the 3x3 boxes and adds two constrained diagonals — see sudoku diagonal.

Hyper sudoku keeps the boxes and adds four extra overlapping 3x3 regions.

Jigsaw with diagonals combines both, and is genuinely difficult.

Samurai jigsaw uses five overlapping jigsaw grids, which is about as far as the format goes before it becomes impractical on paper.

A Worked Example of the Leftovers Rule

It's worth walking through slowly once, because reading the rule and using it are different things.

Picture the top two rows of a jigsaw grid. Region A covers eight cells across those two rows plus one cell that drops into row three. Region B covers the remaining four cells of rows one and two — except that one cell in row two belongs to Region C instead.

Count the cells. Rows one and two hold eighteen cells total. Region A contributes eight of them, Region B contributes four. That's twelve, leaving six cells in those rows belonging to other regions, one of which is the Region C cell we care about.

Now the arithmetic that matters. Regions A and B together contain each digit 1-9 exactly twice — once per region. Rows one and two together also contain each digit exactly twice. The two collections overlap almost completely, differing only by Region A's cell in row three (in the regions but not the rows) and Region C's cell in row two (in the rows but not the regions).

Since both collections hold identical digit counts, and they differ by exactly one cell on each side, those two cells must contain the same digit. If Region A's row-three cell holds a 6, Region C's row-two cell holds a 6 as well.

That's frequently enough to resolve one or both outright, and even when it isn't, knowing two distant cells share a value is a powerful constraint to carry into the rest of the solve.

Building the Habit of Reading Regions

The skill that separates comfortable jigsaw solvers from frustrated ones is almost entirely about grid reading rather than logic. Three things build it:

Trace before solving, every time. Thirty seconds following each region outline with a finger. Skipping this is the most common cause of a solve going wrong forty moves in.

Name the shapes. Solvers who mentally label regions — "the L in the corner," "the long snake," "the fat one in the middle" — attribute cells far more reliably than solvers treating them as an undifferentiated pattern. It sounds childish and it genuinely works.

Check membership before every elimination, early on. For the first several puzzles, consciously verify which region a cell belongs to before erasing a candidate. It's slow and it builds the reflex. After five or six puzzles the checking becomes automatic and the speed returns.

Notice which regions are compact and which are strung out. The compact ones will resolve first and are where your attention belongs in the opening. The strung-out ones resolve last and chasing them early is wasted effort.

Printing Jigsaw Puzzles So the Regions Stay Readable

Jigsaw sudoku is the format most often ruined in printing, because everything about solving it depends on being able to see region boundaries instantly. A standard grid survives a mediocre print — the 3x3 structure is predictable, so you can infer the boxes even when the borders are faint. In a jigsaw grid there is nothing to infer.

Shade the regions, do not rely on borders alone. Alternating light grays across the nine regions is far more readable than heavy outlines, particularly at the points where three regions meet. Four or five distinguishable shades is enough, as long as no two adjacent regions share one.

Check the shading survives your printer. Very light grays often print as white, and dark ones swallow pencil marks. Print one page and test before committing a set — a shade you can write over comfortably in pencil is the target.

Print larger than you would a standard grid. Half-inch cells rather than the usual four-tenths, since jigsaw solving leans harder on complete pencil marks and you will be tracing region outlines with your eyes constantly.

One puzzle per page. Two jigsaw grids side by side read as a single continuous pattern of shapes at a glance, which is exactly the confusion you are trying to avoid.

If a printed set arrives with outlines only and no shading, coloring the regions yourself with a highlighter before starting takes a couple of minutes and repays it several times over.

Frequently Asked Questions

Is jigsaw sudoku the same as squiggly or nonomino sudoku? Yes — jigsaw, squiggly, geometry and nonomino sudoku all describe the same variant: standard sudoku rules with nine irregular nine-cell regions replacing the 3x3 boxes.

Do the regions have to be connected shapes? In virtually all published jigsaw puzzles, yes — each region is a single connected shape of nine cells. Disconnected regions are technically possible but are rare and generally considered poor puzzle design, since they're extremely hard to read.

Does every standard sudoku technique still work? Yes. Scanning, naked and hidden singles, pairs, triples, pointing sets and advanced patterns all transfer directly. Jigsaw adds the law of leftovers on top rather than taking anything away.

Why does jigsaw sudoku feel so much harder than its clue count suggests? Mostly because region membership has to be consciously checked on every deduction, rather than being read automatically the way 3x3 boxes are. That extra lookup slows every step, even though the underlying logic is unchanged.

How long does it take to get comfortable with jigsaw sudoku? Most solvers who already handle standard hard puzzles adjust within five or six jigsaw puzzles. The learning is about reading the grid rather than new logic, which is why it comes fairly quickly once you commit to tracing the regions before starting.

Where can I find a jigsaw sudoku printable? Any generator that supports irregular regions can produce them, and the thing to check before printing a whole set is region shading. Puzzles that mark regions with shading rather than outlines alone are substantially easier to solve on paper, and worth preferring over a source that only draws borders.

Are jigsaw sudoku puzzles harder than standard ones? At the same clue count, generally yes. The regions are irregular, so you cannot scan by habit — every deduction requires reading the region shape first. The law of leftovers partly compensates, but most solvers find a jigsaw puzzle takes longer than a standard grid of the same rating.