Kakuro Combinations: The Number Blocks That Crack the Grid
A run of three white cells with a clue of 6 can be filled exactly one way: 1, 2, and 3, in some order. There is no other set of three different digits from 1–9 that adds to 6. Kakuro is full of these forced runs, and recognising them on sight is the single biggest jump in solving speed you can make.
The rule that makes it work is the one that defines Kakuro: each run must add up to its clue using distinct digits — no digit repeats within a run. That constraint means many clue-and-length pairs have only one possible set of digits, and a few more have only two. Those are the cracks you lever the puzzle open with.
Why combinations matter
A combination tells you which digits live in a run before you know their order. Even when a run has several possible combinations, listing them prunes the candidates hard. And when a run has only one combination, you have placed a full set of constraints on every cell that crosses it — which usually forces digits in the crossing runs immediately.
The "magic blocks" are the runs with a single combination. They are worth memorizing because they appear constantly and they never require any deduction — you just read the clue and the length and write the set.
The blocks worth memorizing
These clue/length pairs have exactly one combination. They are the highest-value patterns to recognise instantly:
- Two cells summing to 3 → 1+2. Summing to 4 → 1+3. Summing to 16 → 7+9. Summing to 17 → 8+9.
- Three cells summing to 6 → 1+2+3. Summing to 7 → 1+2+4. Summing to 23 → 6+8+9. Summing to 24 → 7+8+9.
- Four cells summing to 10 → 1+2+3+4. Summing to 11 → 1+2+3+5. Summing to 29 → 5+7+8+9. Summing to 30 → 6+7+8+9.
Notice the symmetry: the smallest sums force the smallest digits, the largest sums force the largest, and they mirror around the middle. You do not have to memorize a table — once you see that a 3-cell run of 6 must be the three smallest digits and a 3-cell run of 24 must be the three largest, the rest follow from the same idea.
Using a combination to force a crossing run
Suppose a down-run of three cells has a clue of 7, so it must be {1, 2, 4}. One of those cells also sits in an across-run of two with a clue of 17 — which must be {8, 9}. But the shared cell has to satisfy both runs at once, and {1, 2, 4} and {8, 9} share no digits. That is a contradiction, which means you have read the grid wrong somewhere, or one of those clues belongs to a different cell than you thought.
Flip it to the productive case: the down-run is {1, 2, 4} and the crossing across-run of two has a clue of 5, which can be {1, 4} or {2, 3}. The shared cell must be a digit that appears in {1, 2, 4} and in one of those options. {2, 3} contributes only the 2; {1, 4} contributes 1 or 4. Cross-referencing the two runs collapses the possibilities far faster than trying digits one at a time. This is the core loop of Kakuro: combinations narrow a run, crossings narrow it further, and a forced cell sets off the next chain.
Let the board narrow the rest
Most runs have more than one combination, and that is fine — you are not trying to solve a run in isolation. You list its combinations, you list the crossing run's combinations, and you keep only the digits that can appear in both. Often two or three crossings around a single cell are enough to force it, even when no individual run was forced on its own.
In Sukuro, the combination assist does the enumeration for you: select a run and it lists every valid combination for that clue and length, so you can spend your attention on the cross-referencing rather than the arithmetic. On the lower levels it stays visible; higher up it hides behind a tap, so you lean on it less as the patterns become second nature.
Practice it now
Start on a small Kakuro and hunt only for the single-combination runs first — the 3s, 4s, 16s and 17s in two cells, the 6s, 7s, 23s and 24s in three. Fill those, then follow the forced crossings outward. You will be surprised how much of the grid falls out of a handful of magic blocks before you have to do any real deduction at all.