To spot an X-Wing, pick one candidate, find two rows that contain it in exactly the same two columns, then delete that candidate from the rest of those columns. The column version is identical: two columns, same two rows.
The sudoku x wing technique is useful because it turns a visual pattern into a yes-or-no test. You are not asking whether the grid looks promising. You are asking whether, for one digit, two units force the same pair of positions. If they do, one of those two positions must hold the digit in each unit, so every other matching line can lose that candidate.
This is where many players lose time. They scan the whole grid, see a rectangle, and hope it is an X-Wing. Work the other way round. Choose a digit, count its candidate positions inside each row or column, and only then look for matching pairs. In x wing sudoku, the pattern is found by counting first and seeing second.
How do I spot an X-Wing without staring at the whole grid?
- Pick one digit and ignore the other 8. Reason: an X-Wing is always about one candidate, never about a mixture of digits. The wrong move is circling a pretty rectangle made from different candidates; it proves nothing.
- Scan rows first and count candidate positions for that digit. Reason: a useful row has exactly two possible cells for the digit. The wrong move is using a row with 3 or more candidates; the digit is not forced into a pair.
- Write down the column numbers for every two-candidate row. Reason: matching pairs are easier to compare as, for example, columns 2 and 7. The wrong move is trusting your eyes across the grid; box lines and givens make false rectangles look convincing.
- Match two rows with the same two columns. Reason: each of those rows must place the digit in one of the same two columns, so the digit is locked into the corners. The wrong move is matching only one column; that is not an X-Wing.
- Remove the candidate from other cells in those two columns, not from the four corners. Reason: one corner in each row must remain possible. The wrong move is deleting a corner because it looks duplicated; that breaks the logic.
- Repeat the whole check by columns. Reason: column-based X-Wings eliminate along rows. The wrong move is assuming a failed row scan means no X-Wing exists.
What do I check before I mark anything?
Before any deletion, prove the pattern from candidates, not from empty cells. Empty cells are only spaces. Candidates are claims about what can still go there.
- Check that the digit is already excluded from the rest of each base row or column. Reason: the two corners must be the only legal positions in that unit. The wrong move is using two visible pencil marks while a third unmarked legal cell exists.
- Check the two crossing lines contain eliminations outside the corners. Reason: an X-Wing that deletes nothing is real but not useful now. The wrong move is spending time celebrating a pattern that does not change the puzzle.
- Check for singles created before the X-Wing. Reason: naked and hidden singles are cheaper, safer, and often remove the same candidate. The wrong move is forcing an advanced step while a solved cell is waiting.
Then do one final symmetry check. In a row X-Wing, the matching things are columns; in a column X-Wing, the matching things are rows. If you cannot say that sentence aloud for the actual digit, do not mark anything.
Use light pencil discipline. After deleting, look at the affected rows or columns immediately. Reason: the payoff is often a new single or locked candidate. The wrong move is deleting across the grid and then losing the thread of why it was allowed.
When does an X-Wing actually beat simpler moves?
An X-Wing beats simpler moves when the simpler moves are truly exhausted for the candidate involved. The practical test is this: no naked single, hidden single, pointing pair, claiming pair, or obvious locked candidate changes the same area, and one digit has repeated two-position rows or columns. Then the X-Wing is not decoration. It is the shortest legal cut.
Use the sudoku x wing technique when a candidate is scattered enough to block progress, but organised enough to form matching pairs. Reason: singles solve cells; X-Wings clean lines. The wrong move is using it on a digit that already has a direct placement in a row, column, or box. Solve the placement first, because it may destroy or replace the pattern.
It also beats guessing when the puzzle is graded beyond basic tactics. Expert Sudoku puzzles often require X-Wings and harder, so a stall after ordinary scanning is expected. Reason: the puzzle is asking you to remove candidates, not immediately fill a square. The wrong move is treating every stall as permission to search for an X-Wing; some stalls are caused by poor candidate upkeep.
One more rule helps: prefer the X-Wing that removes candidates from a crowded line. Reason: one deletion in a tight column may create a single; three deletions in loose rows may do little. The wrong move is choosing the pattern with the most impressive shape rather than the one with the clearest consequence.
Why does the same pattern sometimes fail to pay off?
- The rectangle uses solved cells, not candidates. Reason: X-Wing logic works on possible positions for one digit. The wrong move is treating four matching givens or placements as if they eliminate anything.
- One base line has a hidden third candidate. Reason: the digit is no longer forced into the two corners. The wrong move is relying on pencil marks that were never updated after a box or column change.
- The eliminations point the wrong way. Reason: row-based X-Wings delete from columns; column-based X-Wings delete from rows. The wrong move is deleting along the two base rows because they look like the main pattern.
- The candidate is already absent from the target cells. Reason: a valid X-Wing can exist without giving a useful deletion. The wrong move is hunting harder and harder for an effect that the grid cannot provide.
- A simpler move changes the structure first. Reason: a single or locked candidate may remove one corner, so the old X-Wing disappears. The wrong move is storing patterns in memory and applying them later after the grid has changed.
- The pattern is actually a larger fish. Reason: Swordfish and similar advanced sudoku techniques can involve 3 base lines, not 2. The wrong move is squeezing a 3-line pattern into an X-Wing and making illegal deletions.
How do I use X-Wings with the other advanced Sudoku techniques?
Treat the X-Wing as a middle step: after routine clean-up, before wider searches that are harder to audit. The table gives a working order, not a law. If a test is decidable and the reason applies, use it; if the wrong-move condition is present, stop.
| Technique | Use with X-Wing | Decidable test | Reason | Wrong move |
|---|---|---|---|---|
| Locked candidates | Check before an X-Wing. | One candidate in a box lies only on 1 row or 1 column. | It gives local eliminations with less risk. | Skipping it and building an X-Wing from candidates it would remove. |
| X-Wing | Use when two lines share the same two positions for one digit. | Exactly 2 base lines, exactly the same 2 crossing lines, same candidate. | It deletes that candidate from the rest of the crossing lines. | Using a base line with a third possible cell. |
| Swordfish | Check after failed X-Wings. | One digit in 3 rows is confined to the same 3 columns, or reversed. | It extends the fish idea when 2 lines are not enough. | Calling it an X-Wing and deleting from only part of the structure. |
| XY-Wing | Use when fish patterns give no deletions. | One bivalue pivot sees 2 bivalue wings that share a removable candidate. | It attacks cells by visibility rather than by parallel lines. | Using it when the pivot or wings have more than 2 candidates. |
| Simple colouring | Use after line-based checks are exhausted. | A single candidate forms a chain of alternating strong links. | It can remove contradictions that no rectangle shows. | Mixing digits in the chain or treating weak links as strong links. |
Frequently asked questions
No. An X-Wing can be row-based or column-based. In a row-based X-Wing, two rows hold the same digit in the same two columns, so you remove that digit from the rest of those columns. In a column-based X-Wing, swap the words row and column. The logic is identical.
4 corner candidates are the visible pattern, but the real count is stricter than that. For one digit, it must appear in exactly two cells in each of two rows, and those cells must sit in the same two columns. Or the same idea with columns first. Extra candidates in the base rows or columns break it.
The common misconception is that any rectangle of four matching candidates is an X-Wing. The correction is that the digit must be locked to two possible positions in each base row or base column. A neat rectangle with spare candidates in the wrong places is only a shape, not a deduction.
Take this board position: for digit 6, row 2 has 6 only in c3 and c8, and row 7 has 6 only in c3 and c8. One of r2c3 and r2c8 must be a 6, and one of r7c3 and r7c8 must be a 6. So no other cell in columns 3 or 8 can be 6.
It depends on the notes: complete, honest candidate marks make X-Wings much easier; patchy notes make them dangerous. If you only write candidates in cells that look promising, you may miss the extra candidate that disproves the pattern. X-Wing is a candidate technique, not a visual decoration.
Sudoku X adds the two main diagonals, so some candidates disappear earlier than they would in standard Sudoku. The X-Wing rule itself does not change: it still uses two rows and two columns for one digit. The difference is that diagonal restrictions may create, or destroy, the exact candidate pattern you need.
Remove only the matching digit outside the four corners, along the two affected lines. The exception is a later deduction that proves one corner false for another reason. The X-Wing itself says the digit must occupy two of the four corners, so deleting a corner candidate is usually deleting part of the proof.
Most people do wrong by chasing rectangles before they have cleaned up singles and simple eliminations. The result is noise: too many candidates, too many almost-patterns, and too much doubt. X-Wings are easier after the grid has been reduced. If the candidate map is messy, fix that before looking for aircraft.
A finned X-Wing is an almost X-Wing with one extra candidate attached in the same box as one corner. That fin weakens the clean four-corner logic, but it can still allow eliminations in cells that see both the fin and the opposite corner. It is harder because the proof is conditional, not symmetric.
Practise one digit at a time. Pick 1, scan all rows for places where 1 appears exactly twice, then check whether any two rows share the same two columns. Repeat for 2, then 3, and so on. Expert Sudoku is useful for this because the puzzles are unlimited and do not require sign-up.
Nobody knows precisely; the best available figure is a personal sample of 20 to 50 completed expert puzzles, counted honestly. Public rates are weak because generators grade puzzles differently, and solvers take different routes through the same grid. On Veena Games, Expert Sudoku is graded by the technique required, so X-Wings belong in the expected toolkit.
In 1977, Star Wars put the X-wing fighter into popular culture. Sudoku writers later used the name for the four-corner pattern whose crossing lines resemble that craft. The evidence for the first Sudoku use of the term is thin, but the name survived because it is easier to remember than a formal row-column candidate lock.
A wrong deletion costs trust. You may still finish the grid, but later contradictions become hard to diagnose because the original mistake is invisible. If you are unsure, do not erase. Mark the suspected X-Wing mentally, check the exact two-per-line condition again, and only then remove candidates from the affected outside cells.
Final thoughts
If you are learning X-Wings, do fewer searches and make each one cleaner. Pick a digit. Count. Record the pairs. Compare. Delete only where the logic points. That rhythm matters more than speed, because speed built on spotting rectangle shapes becomes guessing under pressure.
The useful habit is to name the base lines and target lines before touching a candidate. Say, “7 in rows 3 and 8 is confined to columns 1 and 6, so other 7s leave columns 1 and 6.” If that sentence feels clumsy, the pattern is probably not ready. Leave it alone and look for a simpler move.
Also accept the boring result. Many real X-Wings delete one candidate. Some delete none. That is not failure; it is information about the grid. The mistake is trying to make every found pattern dramatic. Expert solving is often a chain of small, lawful reductions, with an occasional square falling out as payment.
Keep your pencil marks honest. On the next hard grid, test one digit this way before you look for Swordfish, and stop as soon as the count fails.
-
Sudoku
Sudoku X: How Two Diagonals Change the Whole Puzzle
Use the diagonals as extra houses only when they shrink candidates more than the ordinary row, column, or box.
16 August 2026
-
Sudoku
Naked Pairs and Hidden Singles: Sudoku's First Real Techniques
A practical order for using hidden singles and naked pairs, with reasons, traps, and checks you can apply on the next Sudoku grid.
15 August 2026
-
Sudoku
Sudoku Techniques in Order: From First Scan to Expert
Use sudoku techniques in a fixed order: cheap scans first, candidate-based logic next, and expert methods only when simpler tests stop changing the grid.
15 August 2026