THE THREE RULES, RECALLED
Before strategy, the rules: (1) No row or column may contain the same digit twice in unshaded cells. (2) No two shaded cells may be adjacent (horizontally or vertically; diagonal adjacency is fine). (3) All unshaded cells must remain connected as a single group (reachable from each other via horizontal/vertical unshaded neighbours).
Rule 1 tells you WHY a cell gets shaded. Rules 2 and 3 constrain WHICH cells can be shaded. Most Hitori strategy is about exploiting rules 2 and 3 to narrow down which instance of a duplicate is the one that must be shaded.
PATTERN 1: TRIPLE DUPLICATES — SHADE THE MIDDLE
If the same digit appears three times in a row (or column), the middle cell must be shaded. Shade either of the two outer cells and you create adjacent shaded cells — which Rule 2 forbids. The middle is the only valid shade.
This is the most reliable forced pattern in Hitori and should be the first pass on any grid. Look for runs of three identical digits in the same row or column and shade the middle one immediately.
Consequence: once the middle is shaded, the two outer cells are confirmed unshaded — they now block further shading in their rows and columns. Use them immediately to check for new duplicates they create.
PATTERN 2: ADJACENT DUPLICATES — SHADE AWAY FROM ADJACENCY
When two identical digits are adjacent (directly next to each other), exactly one must be shaded. But Rule 2 means the cell adjacent to a shaded cell cannot also be shaded. This limits which instance can be chosen.
If one of the two adjacent duplicates is itself adjacent to a cell that is already shaded (or will be forced shaded by another rule), shading that cell would violate Rule 2. Therefore the OTHER cell must be shaded.
Even without a pre-shaded neighbour, adjacent duplicates give you information: whichever one you shade, its other neighbours become confirmed unshaded (no two shaded can be adjacent). Propagate those confirmations — they often trigger new forced shades elsewhere.
PATTERN 3: THE CONNECTIVITY CONSTRAINT AS A POSITIVE TOOL
Most beginners treat Rule 3 (all unshaded cells connected) as a final check: solve the grid, verify connectivity, done. The real use is as a forward-inference tool that rules out shades.
If shading a particular cell would disconnect the unshaded region — creating an isolated group of cells — that shade is forbidden, regardless of what the duplicates rule suggests. You can use this to confirm the other instance of a duplicate must be shaded.
The practical technique: look for cells that would act as "bridges" — the only unshaded connection between two parts of the grid. A bridge cell cannot be shaded. If a bridge cell happens to contain a duplicate digit, the OTHER instance of that digit must be shaded.
Corner cells and cells at the edge of the grid are particularly valuable bridges to check — they have fewer unshaded neighbours, so removing them is more likely to disconnect a region.
PATTERN 4: CONFIRMED-UNSHADED PROPAGATION
Every time you confirm a cell unshaded (for any reason — it's adjacent to a shade, it's the only unique instance in its row, it's a bridge), check whether it now creates new duplicates in its row and column.
A confirmed-unshaded cell locks in a digit in its row and column. Any other instance of that digit in the same row or column must now be shaded. This is the most common cascade: one shade forces an unshade (Rule 2), which forces a shade (Rule 1 via confirmed-unshaded propagation), which forces another unshade, and so on.
Getting fluent at confirmed-unshaded propagation is what separates fast Hitori solvers from slow ones. After every shade and every confirmed-unshaded deduction, do a full row + column scan for the digit involved before moving on.
PATTERN 5: SANDWICH PATTERNS
A sandwich is when a cell is flanked on both sides (in the same row or column) by two cells that are already confirmed shaded. Since shaded cells cannot be adjacent, the central cell between two shades must be unshaded. This is a direct application of Rule 2 and can feel surprising if you haven't seen it before.
The reverse sandwich is even more useful: if a cell must be shaded (it's the only remaining instance of a duplicate), and shading it would create two adjacent shades, then one of the adjacent cells was wrongly assumed shaded — backtrack or constrain.
Sandwiches appear naturally as a grid fills in. Keep an eye on developing "shade chains" — sequences of shaded cells near each other — and check what they force the cells between and around them to be.
HANDLING AMBIGUITY: CONTRADICTION ANALYSIS
On hard Hitori grids, the five patterns above will stall before the puzzle is solved. The remaining cells have two valid shade assignments that both satisfy the three rules. This is where contradiction analysis takes over.
Pick the ambiguous cell that is most constrained — ideally one that creates the most immediate consequences in either direction. Assume it is shaded and trace the forced deductions: does the chain produce a contradiction (adjacent shades, isolated region, impossible duplicate)? If yes, the cell must be unshaded.
If the "shaded" assumption reaches the same forced deduction as the "unshaded" assumption — i.e., both paths force a third cell to be shaded — that third cell is shaded regardless. This is a "shared forced deduction" and is always sound.
Well-constructed Hitori puzzles (including all GridJoy puzzles) have a unique solution reachable by logic. Contradiction analysis is always sufficient — you should never need to guess and backtrack repeatedly.
THE SOLVING ORDER THAT WORKS
- First pass: Triple duplicates — shade middles, confirm outers unshaded.
- Second pass: Adjacent duplicates constrained by existing shades — shade away from forced adjacency.
- Third pass:Confirmed-unshaded propagation — each new confirmation locks a digit and may force a shade.
- Fourth pass:Connectivity check — identify bridge cells and confirm them unshaded.
- Repeat until stalled.
- If stalled:Contradiction analysis on the most constrained ambiguous cell.
Most beginner and intermediate Hitori grids resolve entirely in the first three passes. The connectivity check and contradiction analysis are tools for harder grids — but having them in mind makes the easier passes more efficient, because you'll notice bridge candidates naturally as you scan.