Dots and boxes is the most misunderstood of all the paper games. Casual players believe it's about grabbing boxes as fast as possible. It isn't. Between two players who know the deeper game, almost every box changes hands in a few huge bursts at the end — and the winner is whoever controlled when those bursts happened. Learn two ideas — chains and the double-cross — and you will beat greedy players almost every time.
Phase one: the polite opening
Early on, both players add lines while avoiding giving anything away. The rule is simple: never draw the third side of a box unless you're taking the fourth. A box with three sides is a gift — your opponent completes it, scores, and moves again (completing a box always earns another turn).
Count "safe" moves — lines that don't create a three-sided box anywhere. The opening is essentially a countdown: safe moves run out, and whoever runs out first has to open a region. So the real opening skill is shaping where the unsafe moves will be while burning as few safe moves as you can.
Phase two: chains decide everything
As the grid fills, the empty boxes organise themselves into chains — strings of adjacent boxes that will all fall together once one of them is opened. Give your opponent the first box of a five-box chain and, because each completed box grants another move, they take all five.
This leads to the single most important idea in the game, known as the chain rule: you want the number of long chains to have the right parity. As a practical rule of thumb for the standard game: the first player wants the total number of long chains (3+ boxes) to be odd; the second player wants it even. During the neutral phase, you're not just avoiding gifts — you're fighting to merge or split chains so the count ends up on your side. That's the invisible battle casual players never see.
Phase three: the double-cross
Here's the move that wins games outright. Your opponent is forced to open a chain of, say, six boxes. The greedy play is to take all six. The winning play is usually to take four and then — instead of taking the last two — draw the line that leaves both of them for your opponent, as a single "domino".
Why give away two boxes? Because taking the domino forces your opponent to move again — and the only moves left are opening the next chain, for you. You paid two boxes to stay in control. Repeat this at the end of every chain and your opponent opens every remaining chain on the board while you harvest them. Two boxes each time buys you chains of five, seven, nine… It feels rude the first time you do it. It is also completely devastating.
Sacrifices earlier in the game
The same logic applies before the endgame: sometimes deliberately giving your opponent a short chain (two boxes, ideally) is correct, because it forces them to open a long chain next. When you must open something, always open the shortest available region — and check whether a two-box sacrifice now converts into a seven-box harvest later.
Practical checklist
- Early: never create a third side; count remaining safe moves out loud (in your head).
- Middle: count the chains forming. First player fights for an odd number of long chains, second for even.
- Someone must break first — make sure it's them, or open the smallest region if it's you.
- Endgame: all-but-two. Take every box in a chain except the last two; leave the domino.
- Exception: take everything in the final chain of the game — there's nothing left to control.
Board size notes
On RivalPit you can play from 3×3 up to 6×6 boxes. Small boards (3×3) are almost pure tactics — chain parity is decided within the first few moves, so study them to drill the endgame. Bigger boards (5×5, 6×6) give the chain-shaping battle room to breathe and are where the double-cross really shines, with three or four long chains regularly changing hands in the final minute.
The computer opponent here plays a solid greedy-plus game — it takes what's offered and avoids obvious gifts — which makes it the perfect training partner for practising your double-cross discipline before you unleash it on a human.