Not every game ends in a win or a loss. When neither side can force a result, or both sides accept that neither can, the game is drawn and each player scores half a point — the same half point you'd get for a 3-3 tie after being down 3-0 at halftime: not a win, but sometimes exactly the result you were hoping for. Chess played with perfect accuracy on both sides is theoretically a draw. Pit two top engines like Stockfish and Komodo against each other in a 100-game match with matching hardware, and it would be no surprise if every single game were drawn. That's the whole reason engine-vs-engine tournaments like the Computer Chess Championship don't just let two top engines play freely from the start: the organizers force each pairing through a variety of predetermined, slightly unbalanced opening positions (playing the pairing twice, with colors reversed, in each one), specifically to manufacture chances for a decisive result. Left alone, two engines of matched strength will simply draw again and again.
Human games are far less clean, and that gap between "theoretically a draw" and "actually played by people" is what makes the rest of this chapter useful: a draw can be the best possible outcome from a lost position, and a real threat you have to watch for when you're the one who's winning.
Fun fact: across roughly 2.4 million master games on the Lichess database, about 43% end in draws, White wins around 33%, and Black wins around 24%. Zoom out to around 220 million games from ordinary club-level players (roughly 1600-rated) and the picture flips completely: only about 4% end in draws, while White wins about 50% and Black about 46%. Even an International Master with a peak rating of 2430 — Rozman himself — will tell you he still blunders plenty; for the rest of us, chess is nowhere near a solved, drawish game in practice, which is exactly why knowing how to grab a draw out of a bad position, or avoid gifting one away from a winning one, matters so much below master level.
Either player can offer a draw at any point, and the opponent is free to accept or decline. This makes sense in genuinely balanced positions, where any attempt to force a win carries more risk than reward against a comparably skilled opponent.
Draw agreements don't require perfect symmetry, either. Here's a real one, from the last round of the 2022 Chess Olympiad.
At the top level, agreed draws have become controversial: unlike almost any other competitive sport, chess lets players end a contest early by mutual consent. Some tournaments (Sofia, Bulgaria, in 2005, was an early example) have banned draw offers outright to encourage more fighting chess — the thinking being that even strong grandmasters, especially with the Black pieces, can get too comfortable settling for a safe result. At the beginner and intermediate level this is rarely an issue: as the stats above show, players below master strength are already fighting to the end 96% of the time, offer or no offer.
If the same position occurs three times — not necessarily in a row, and not necessarily reached by the same sequence of moves — the game is a draw. This most often comes up as perpetual check: a defending side that's losing badly on material can sometimes save the game by checking the enemy king over and over, if the opponent's king can never escape the checks or block them permanently. The position below looks completely lost for White — see if you can find the drawing resource before you check the solution.
Repetition doesn't need a queen, either — any piece that can keep giving check while the king has only one reply works the same way. This next one grew out of a Pirc Defense where White had just won Black's queen outright — a dream result on the scoreboard, or so it seems. See if you can find Black's saving resource before you check the solution.
Repetition cuts both ways. If you're much worse, hunt for a perpetual check. If you're much better, always double-check that your opponent doesn't have one hiding in the position before you relax.
Stalemate is when the player to move has no legal moves at all but is not in check. It's one of the most heartbreaking ways to draw a game, because it typically happens to a player who is completely winning and simply mishandles the finish.
Stalemate isn't limited to a cramped corner, either. Being up even more material doesn't protect you if every escape square is covered without check.
Stalemate can also strike with a whole army still on the board, not just a bare king. This is a real position from one of Rozman's own blitz games, against his good friend and fellow International Master Eric Rosen: he had just captured his opponent's queen with the rook on f8, up 13 points of material with a completely dominant position.
The fix, in hindsight, was simple: instead of grabbing the free queen immediately, the author could have inserted a check first — the queen from e2 to h5 checks the Black king on h8 — and only then captured the queen, which would have been checkmate rather than a gift. The lesson generalizes: whenever every one of your opponent's pieces is nearly frozen, ask whether a check is available before you cash in the material.
When you're winning big, slow down near the finish. Ask "does my opponent have any legal move at all?" before every move — especially before a capture that looks free. If the answer would be no, and your move doesn't give check, you may be about to hand back the whole game.
Stalemate is genuinely rare at grandmaster level — strong players are too experienced to blunder into one. It's overwhelmingly a beginner-and-intermediate problem, which is exactly why the checklist above is worth making into a habit early.
If 50 moves pass for each player with no capture and no pawn move, the game is automatically drawn. Without this rule, a player with an insufficient or merely symbolic advantage could shuffle pieces forever, hoping the opponent eventually blunders, in a game with no natural end.
The rule doesn't stop a side that's genuinely making progress — captures and pawn moves reset the counter — it only kills games where nothing is actually changing.
Some material combinations can never force checkmate against a lone king, however well they're played. When both sides are reduced to such material, the game ends immediately in a draw — no mistake required from either side.
A lone pawn is different, even though it's worth less than a knight or bishop:
Insufficient material can also rescue a losing position outright — if you're down a rook but manage to trade down to king and bishop against a bare king, the game is a draw regardless of what happened earlier. It even overrides the clock. In the position below, Black is completely winning and has more than one way to deliver mate immediately — see how many you can find before you check the solution.
Chess offers five distinct roads to a draw: agreement between two players who see no way forward, threefold repetition (often via perpetual check when one side is losing), stalemate (no legal moves and no check — a real danger when you're up material and careless), the 50-move rule (protecting stale positions with no captures or pawn moves), and insufficient material (some combinations of pieces simply can't force mate). Knowing these well cuts both ways: they're tools to save a lost game, and traps to avoid when you're winning one. The practical lesson for a beginner is simple — never resign. Play the position out, because your opponent might blunder into a stalemate, get stuck repeating moves, or run out of time before you do.
Agreement, threefold repetition, stalemate, the 50-move rule, and insufficient checkmating material.
Repeatedly checking the enemy king so it can never escape or block permanently, forcing the position to repeat. It's most useful when you're losing badly on material — it turns a loss into a draw.
The player to move has no legal move and is not in check. It's painful because it typically happens to a player with a large material advantage who simply fails to leave the opposing king any legal move while forgetting to check it.
Because the Black king, although completely boxed in by the two queens, is not currently in check. Having no legal moves while not in check is the definition of stalemate, not checkmate — material size is irrelevant to that distinction.
It exists so that a side with a merely nominal or unconvertable advantage can't shuffle pieces forever hoping for a blunder. Any capture or any pawn move resets the count back to zero.
No — a lone pawn is never insufficient material, even though it's worth less than a knight or bishop, because it can promote into a queen or rook if it's safely escorted to the far rank.
A draw, not a win. Because your material could never have forced checkmate regardless of the clock, insufficient material overrides the normal "loses on time" rule.
Rook versus knight with no pawns is a well-known theoretical draw, so you likely can't force a win with accurate defense from the other side — and even if you keep trying, the 50-move rule will eventually draw the game if you make no real progress.
Because chess played with perfect accuracy is theoretically a draw, so two closely matched engines starting from the normal position would simply draw almost every game. Forcing predetermined, slightly imbalanced starting positions (and playing each one twice with colors reversed) manufactures the winning and losing chances that pure best play would otherwise erase.
Around 43% of master games end in draws (White winning about 33%, Black about 24%), versus only about 4% of average club-level games (White about 50%, Black about 46%). The gap exists because stronger players convert small, well-defended advantages into draws far more reliably, while weaker players tend to keep making mistakes until someone wins outright.
A check first — moving the queen from e2 to h5 checks the king on h8. Only the knight can legally block on h6, after which capturing the queen on f8 delivers checkmate along the back rank instead of stalemate. The general lesson: when your opponent's pieces are nearly frozen, check for an available check before you cash in a "free" capture.