XY-Wing Sudoku Technique Explained for 2026

XY-Wing is a three-cell pattern built entirely from bivalue cells—cells that still hold exactly two candidates. Unlike line-based fish (X-Wing, Swordfish), XY-Wing lives in the geometry of visibility: one cell shares a unit with each of the other two, while those two “pincers” do not see each other. That asymmetry is what forces a digit to behave like a remote pair and produce a clean elimination.

Use XY-Wing when the grid is rich in pairs but no conjugate track gives you a strong line pattern yet; it is especially common once boxes start locking candidates but rows and columns still look opaque.

Pattern

Label the three candidates involved as X, Y, and Z (all different digits). The structure is:

1. Hinge (pivot): a bivalue cell {X,Y}.
2. Pincer A: bivalue {X,Z} and lies in a row, column, or box with the hinge (so it sees the hinge).
3. Pincer B: bivalue {Y,Z} and lies in a different row/column/box overlap with the hinge (so it also sees the hinge).
4. Key constraint: the two pincers must not see each other.

You will often spot it as a small triangle of {a,b} cells where two pairs share a digit with the hinge but the outer cells cannot see each other.

Logic

Why the elimination follows: the hinge must be X or Y. If it is X, pincer A collapses toward Z; if it is Y, pincer B collapses toward Z. Either way, Z must appear on at least one pincer in every completion of the pattern. Therefore any cell that sees both pincers cannot still claim Z as a candidate.

Reach for XY-Wing when:

• You have multiple bivalue cells clustered around one or two boxes.
• X-Wing / Swordfish scans return nothing, but pairs keep “pointing” at the same digit from different directions.
• After intersection work (pointing, claiming), several {a,b} pairs remain in a tight neighborhood.

XY-Wing is a local pattern: it does not require a full-row conjugate count like fish. That makes it a natural bridge between intermediate pair tactics and harder net techniques.

Example

This is a logic scaffold you can map onto your own grids; the point is the case split on the hinge, not a specific puzzle copy.

Suppose:

• Hinge at R5C5 is {3,7}.
• Pincer A at R5C8 is {3,9} (same row as the hinge → sees the hinge).
• Pincer B at R8C5 is {7,9} (same column as the hinge → sees the hinge).
• R8C8 sees both R5C8 and R8C5.

Case 1 — hinge is 3: then R5C8 cannot be 3, so R5C8 resolves toward 9.

Case 2 — hinge is 7: then R8C5 cannot be 7, so R8C5 resolves toward 9.

In either case, a 9 is forced onto at least one of the pincers, and R8C8 sees both. Eliminate 9 from R8C8.

Diagram (relationships, not a full puzzle):

        col 5    col 8
row 5   [3,7]----[3,9]   <- hinge shares row with pincer A
          |
row 8   [7,9]      ?    <- pincer B shares column with hinge
                 (R8C8 sees both pincers -> loses 9)

After you remove the candidate, scan for naked singles or a follow-up XY-Wing with a new hinge—hard puzzles often chain several small wings before fish reappear.

Next step: Sudoku Face Off keeps pencil marks disciplined so hinge cells stay readable as the puzzle tightens.

Pitfalls

• False wings: if either “pincer” has more than two candidates left, it is not a legal XY-Wing node.
• Visibility mistakes: pincers must see the hinge, but must not see each other; mixing this up is the #1 silent failure mode.
• Digit drift: the shared digit between hinge and each pincer must line up exactly ({X,Y} with {X,Z} and {Y,Z}); a near-miss often means you are looking at a UR or W-Wing instead.

For line-based fish, continue to X-Wing and Swordfish, or the combined X-Wing and Swordfish guide. For a wider map of essentials, read essential advanced Sudoku techniques—then return here when you want hinge logic spelled out on its own URL.