How to play Star Grid

Place one star in every row, column and colour, and never let two of them touch. Each grid is proved to have exactly one answer before you ever see it.

Play Star Grid free → More puzzle games

Star Battle — the puzzle this game is built on — is pure deduction. There is no luck in it, no hidden information, and in this version no guessing either. Everything below follows from three rules and one guarantee.

The three rules

Every row contains exactly one star. Every column contains exactly one star. Every coloured region contains exactly one star. On a grid ten squares across that means ten stars, no more and no fewer, and each of the thirty constraints is satisfied exactly rather than approximately.

No two stars may touch. Not side by side, not one above the other, and — the part people forget — not diagonally either. A placed star therefore rules out its entire row, its entire column, its entire colour, and the ring of up to eight squares surrounding it.

The regions are irregular and they are the whole difficulty. A region may be a compact blob, a long snake winding across the grid, or a single square. A single-square region is worth spotting immediately: it is a star given to you free, and everything it rules out comes free with it.

Why the touching rule matters more than it looks

It is tempting to read "no two stars adjacent" as a tie-breaker and concentrate on the rows and columns. That gets it backwards. One star per row and one per column is a permutation, and permutations are plentiful — a grid with those two rules alone would have an enormous number of answers. The touching rule and the regions are what cut that down to one.

In practice the touching rule is what makes a placement pay. Committing a star does not just tick off a row and a column; it takes three squares out of the row above and three out of the row below, and on a small grid that is often enough to strand a colour or force a neighbour. When you are hunting for the next deduction, look at what a candidate would destroy rather than what it would satisfy.

What the difficulty labels mean

Not grid size. Every grid is rated by the hardest technique it genuinely forces, and "forces" is tested the hard way — the technique is switched off and the grid is re-solved to see whether it still falls. A technique that merely fires along the way does not raise the rating.

Easy grids are settled by the last square left. Work through the fallout of the stars you have already placed and eventually some row, column or colour has exactly one square standing, so that square takes a star. Nothing more is required.

Medium grids force confinement. This is the first real technique: if every square a colour has left lies on one row, then that row's star belongs to that colour, so every other square on that row can be crossed out even though you do not yet know which square the star occupies. The argument runs in both directions — a row whose remaining squares all sit inside one colour tells you the same thing about that colour.

Hard grids force the same argument on several units at once. If three colours between them have candidates on only three rows, those three rows belong to those three colours, and every other colour is shut out of them. It is confinement generalised, and it is the point where the puzzle starts to feel like real deduction.

Expert grids cannot be finished by any of that alone. At some point you must assume a star sits on a particular square, follow the consequences until something breaks — a row with no squares left, or a colour that can no longer be satisfied — and conclude the assumption was wrong. It is still deduction and the contradiction is always genuinely reachable; it is simply one step deeper.

The guarantee

Two things are proved about every grid before it is dealt. First, exactly one arrangement of stars satisfies it. Second, a solver restricted to the techniques above can reach that arrangement. Any grid failing either check is discarded and regenerated.

So if you are stuck, something is deducible. You have not been handed a grid that needs a coin flip, and there is no point at which two answers remain and you must pick one.

Hints, undo and the daily grid

A hint reads the board as you have actually left it. Your pencil marks are deliberately ignored, because advice built on top of a square you crossed out by mistake would be worth nothing — but your stars are trusted, so the hint follows the line you are actually on. If one of those stars cannot be part of any answer, the hint says so and marks it, instead of helping you deeper into a dead end.

Undo steps back one mark. Reset clears your marks without changing the grid.

One grid a day, the same for everybody. The week ramps: Monday and Tuesday are easy, Wednesday and Thursday medium, Friday and Saturday hard, and Sunday is an expert grid. Solving extends your streak, solves finished without a single hint are counted separately, and the share text carries your time without giving anything away. You can also send the exact grid you just played to someone else, which deals them your board rather than a fresh one.

Controls

Tap a square once to pencil it out, again to place a star, and once more to clear it. Dragging across empty squares pencils out a whole run of them at once.

Keyboard

Arrow keys move the cursor around the grid.
Space or Enter cycles the square under the cursor.
X pencils a square out, and back in again.
Z undoes, H asks for a hint, R resets the grid.

Strategy tips

Find the smallest colours first. A colour that is a single square is a free star, and a colour of two or three squares is usually the next thing to fall.
Work the corners and edges early. A star in a corner poisons only three neighbouring squares instead of eight, so corner regions constrain the grid less and tend to resolve late — which means the middle is where your effort pays.
After placing a star, cross out its ring before anything else. The three squares above and the three below are the deductions people leave on the table, and they are free.
When counting stops working, look for a colour trapped on one line. If every square a colour has left is on one row, that row belongs to it and the rest of the row can go, even though you cannot yet say which square holds the star.
Run that argument backwards too. A row whose remaining squares all sit in one colour tells you that colour's star is on the row, so the colour's squares elsewhere can be crossed out.
On the harder grids, count colours against lines. If three colours have candidates on only three rows between them, those rows are spoken for and every other colour is shut out of them.
Use the pencil marks properly. Crossing out squares you have ruled out is not bookkeeping — it is what makes a confined colour visible in the first place, and solvers who skip it stall on grids they could have finished.
When everything is exhausted on an expert grid, pick a square with only two plausible homes and assume one. Follow it until a row empties or a colour dies, then rule it out. The contradiction is always there to be found.