Garry Kasparov. My Best Chess Game
Article from the magazine Chess In Russia, 1999
My Best Chess Game
"My best game is ahead..." How many times have we heard this answer from chess players of any age and level of success. The dream of playing that game, which would shine upon the player's legacy until the end of their active career, no matter how many other brilliant games they played and how many other deep, original combinations they performed - still, the dream of something non-achievable is the thing that drives chess players to explore new depths of the game. But I'm afraid that after January 20th 1999, I'd have a hard time to convince anybody that my best game is still ahead. The notion of "best game" is very subjective: some players like combinations, others - clear positional games, still others have different criteria. There is no universal, one-size-fits-all definition, because any chess player has their own preferences regarding perfection in chess.
Among my games, there are several that can satisfy even the strictest connoisseurs, but only a few can be named as contenders for that game. Two of them, by the way, wound up in the all-time top 3 games of Chess Informant. The first is against Karpov (16th in the 1985 match), the other is against Anand (10th in the 1995 match). And still, these games had their flaws: an objectively incorrect novelty (pawn sacrifice in the opening against Karpov), and against Anand, the integrity and beauty of the game are still compromised by the fact that most of it was prepared at home. But I think that the main criterion for beauty is the game's impression on others.
It's very hard to describe how should a dazzling combination look, especially now, at the end of 20th century, where both amateurs and grandmasters can use computers for help. Any combination, any sacrifice can be scrutinized not only by thorough analysts, but also by any amateur chess player who can buy a powerful enough computer and install a strong chess program. So, today's criterion of beauty is both integrity and correctness of the intention, and, of course, some human flaws, because it always takes two to create a combination. Everyone now can assess the correctness of any combination much quicker. We don't need years or months now - the verdict is out in mere days or weeks. It's obvious that the combination must be mind-blowing, with mating final, with sacrifices, where mind prevails over matter.
This should be a mating combination where a mating net is created by a very small force. Everyone likes that. This is, after all, the goal of a chess game - to checkmate the opposing King. But modern defence technique doesn't allow for such combinations, nips them in the bud. Piece or pawn sacrifices have became rare, and a Rook sacrifice without obvious combinational motives is something of a relic on the highest level. The legendary Kings' marches along the whole board, dodging checks all the way, are now also the thing of the distant past. The times of Anderssen's Immortal and Evergreen game have long passed. And I've never thought that I'd ever be able to resurrect this rebellious, romantic spirit on the chess board and create a combination that fits all possible criteria.
But let's wrap up the foreword. It's 4th round, 20th January 1999. In a small, windy Dutch village named Wijk-an-Zee, in a huge hall by the sea, a big chess family gathered: the world's leading chess players, the B-tournament grandmasters, participants of various open tournaments. I actually became a stranger to such conditions in recent years. The day didn't promise much. Yes, we had a blitz tournament where I won 7 games in a row (including a good endgame victory over Yermolinsky), so I was in a good mood before the Topalov game. Now it's hard to say whether I expected something special, though I did have some strange premonitions. I had similar premonitions before: I remember my feelings before the game against Karpov in Linares and before the game against Gelfand in the next round. I won both games brilliantly, but here, the feelings were different.
Garry Kasparov - Veselin Topalov, Wijk-an-Zee 1999
Nothing portended a storm after I made this move. Topalov, who's always ready to fight, no matter whom he plays and with which colour, played
Frankly, I was surprised. The Ufimtsev defence isn't a part of Topalov's repertoire and, in my opinion, isn't an opening worthy of using in top-level tournaments. White have too many opportunities. And very diverse, too: they can play sharply, or positionally, use various methods of getting the initiative. Nevertheless, Topalov seemed to expect that I'd be surprised and play worse in an unknown situation, besides, he wanted to avoid my opening preparations.
2. d4 Nf6 3. Nc3 g6
Here, I started thinking. Actually, I already started thinking on the third move: I've often played 3. f3, threatening to transpose into King's Indian, but Topalov isn't someone who can be threatened with this opening, and he was probably expecting it. So I've decided to make it up as I went along, to get a position that I had some idea about, even though I've never had it in my practice before and never analyzed it seriously.
4. Be3 Bg7 5. Qd2 c6
As far as I know, Black usually play c6 and b5 before Bg7, but I don't think that the move order affects anything here.
Also possible was 6. Nf3 b5 7. Bd3, perhaps it was more precise, but, as I said, both opponents played the opening based on common sense, not on preparation.
6... b5 7. Nge2
A strange move. If White wanted to play 7. Bh6!?, they could have done it immediately, leaving the e2 square for the other Knight and retaining the opportunity to develop their Bishop to d3. This Knight could have theoretically gone to h3 at some point. The move 7. Nge2 has no particular meaning, except for psychological. I remembered that before the game, when we discussed the strategy, Yuri Dokhoyan who looked through some Topalov games said suddenly: "You know, Garry, he doesn't like when his partner makes moves that he couldn't predict. This affects him in strange ways." Upon remembering the phrase, I played 7. Nge2, and Topalov was indeed surprised. This move doesn't carry any threats, White just continue their development. But it seemed to me that he didn't like the game's overall development - he probably intended something very different before the game.
7... Nbd7 8. Bh6
Better late than never. It's useful to exchange the Bishop.
8... Bxh6 9. Qxh6
White have some limited success: Black can't castle short now. But this achievement doesn't mean much because the Black King can escape to queenside, where the White King will also end up, and so, a quiet, maneuvering game lies ahead, without much advantage for White. If Black try to show activity with 9... Qa5, White have 10. Nc1 with subsequent Nb3. White will stabilize the game and take away the opportunity to exploit their Queen's distraction.
9... Bb7 10. a3
I didn't want to castle immediately, because after 10... Qa5, it's unclear how to defend against b4, so White make a quiet move preparing to castle long. Now 10... Qa5 can again met with 11. Nc1, preventing b4.
Topalov, after much thinking (11 minutes), decides to fortify the center and castle long. Black had other plans too, but this one was the most logical.
11. O-O-O Qe7 12. Kb1
White don't have too many opportunities too: they have to untangle their pieces, and so I planned to get my Knight to b3. If Black try to play actively, 12... a5?! is met with 13. Nc1 b4 14. dxe5! dxe5 (14... Ng8 15. Qg7 Qxe5 16. Qxe5 dxe5 17. Na4 +-) 15. Na4 bxa3 16. b3 +-.
Black could castle immediately, but Topalov, just in case, prevents d5. I doubt that this was a real threat, but still, the move was more or less necessary.
13. Nc1 O-O-O 14. Nb3
The development has ended, but now Black have to be enterprising, because their position is cramped. If White end their development, say, with g3, Bh3 and Rhe1, Black would have a hard time. The Black King is somewhat weak, and they could have planned c5, but then White would have a choice: either close the position with d5 or exchange pawns. Closing the center would be more promising. The space advantage could have helped to create an attack. Here, I had an idea to exploit Black's queenside weaknesses. I could move my Queen from h6 to b6 or a7. This ridiculous thought came and went, but somewhere deep in my subconscious, the idea that Queen at b6 and Knight at a5 could wreak havoc, especially with the Bishop at h3, took hold. This didn't affect my calculations, but the idea itself was the prologue for a great combination.
A very good decision: decreasing the tension in the center. Black, using the fact that White are somewhat behind in development, don't hesitate to open up the play, believing that their active pieces would compensate the King's weakened position.
15. Rxd4 c5 16. Rd1 Nb6!
A good move. Black are preparing d5, and I had to think for 10 minutes here. What White should do now? For instance, after 17. a4?!, Black get a good position with 17... b4 18. a5 bxc3 19. axb6 Nd7. And after 17. Na5 d5 18. Nxb7 (18. g3 d4) 18... Kxb7 19. exd5 Nbxd5 20. Nxd5 Nxd5 21. Bd3 f5 22. Rhe1 Qc7 23. Bf1 c4, a complicated position with mutual chances arose. The Black King is, of course, exposed, but the White Bishop is limited by pawns, so the position is roughly equal.
The contours of combinations have already started to appear. I didn't clearly see it yet, but I knew that g3 and Bh3 wouldn't be bad.
Now the Bishop will go to h3, the Queen will retreat to f4, the Knight will go to a5, and there will be some kind of attack, I didn't know what kind yet... But I was already fixated on the idea to position my pieces like that.
Topalov thinks that he has enough time to prepare d5.
18. Na5 Ba8 19. Bh3 d5
Both opponents have completed their plans. White finished their development, and Black played d6-d5. It's important to notice that if White first played Bh3, then their Knight wouldn't go to a5 due to Nc4. Though I could still play Rhe1 after that, but that game would have been different. I was almost automatically playing out my plan for a sacrifice. I knew I would play Rxd4, even though I haven't even seen the draw by repetition yet.
20. Qf4+ Ka7 21. Rhe1
Now, I did see the draw by repetition. Moreover, I felt that it was possible to play on without a Rook, even though I couldn't see what it led to. But the vision of the Black King at a4 warmed up my soul, and the innate "attacker's intuition" (let's call it that) that all humans have got told me that there is some solution, and the mating net is indeed possible, despite the opponent's huge material advantage. Also, I was driven by curiosity, the feeling of unknown. When else would I be able to create something like that, to draw the Black King deep into my own territory? We all know that legendary Edward Lasker game with the sacrifice at h7 and Black King's march from g8 to g1. This could happen only in a bygone era. But now, this opportunity! Topalov looked confident. He played
Of course, after 21... dxe4? 22. fxe4 the game opens up, and the threat of 23. Nd5 gives Black a hard time: their King is too weak. White, of course, could have played 22. Na2, but after 22... Rhe8 or 22... h6, there was a difficult game ahead. But my hand already put my Knight at the central square.
Objectively, this move isn't strongest, but it leads to the following combination.
22... Nbxd5 23. exd5 Qd6
Topalov looked somewhat bemused because he seemed to think that White's attack has fizzled out. There's no sense to give a check at c6: the Knight will be taken, then the King goes to b6, and White Rooks can't get there. The d4 pawn blocks the d-file, and there's nowhere to invade at the e-file. But it's not quite like that, and my next move became an unpleasant surprise for Topalov.
When I made this move, I saw only the draw by repetition and an opportunity to continue the attack, but all the pieces of the variant weren't there yet. I did see the idea of Rd6-Rb6, but still I thought that the variants should have been checked in full. Perhaps Black would find some hidden defence resources. Topalov thought for 15 or so minutes. I paced along the hall, and during this frantic pacing, I got an impression that most games were already over, and the players left. My thoughts worked in one direction, and then I saw it, the entire branch of variants. I saw Rd7, I don't even remember how it all came together, but I saw the entire line, saw the Black King's passage after Bf1, saw Rd7, and couldn't hold my excitement, because then I immediately saw that 24... Kb6 destroys everything.
I was going to play 25. b4, underestimating Black's better position after 25... Qxf4 (25... Nxd5 26 Qxd6+ Rxd6 27. bxc5+ Kxc5 28. Nb3+ Kc6 29. Kb2 Rhd8 30. Red1 Bc6 31. f4 Kc7 =) 26. Rxf4 Nxd5 27. Rxf7 cxb4 28. axb4 Nxb4 29. Nb3 Rd6. Perhaps if Topalov did play 24... Kb6! (24... Bxd5?! 25. Rxd5! Nxd5 26. Qxf7+ Nc7 27. Re6 Rd7 28. Rxd6 Rxf7 29. Nc6+ Ka8 30. f4, White are slightly better), I would have found 25. Nb3!, after which Black again couldn't take the Rook: 25... Bxd5! (25... cxd4? 26. Qxd4+ Kc7 27. Qa7+ Bb7 28. Nc5 Rb8 29. Re7+ +-; 25... Nxd5? 26. Qxf7 Rhf8 27. Qg7 Rg8 28. Qh6 Qf8 29. Rh4 +-) 26. Qxd6+ Rxd6 27. Rd2 Rhd8 28. Red1, and White are equal at best. The very thought of this combination possibly failing was nerve-wracking, and I prayed that Topalov would capture at d4. I still wasn't sure that this won, but the beauty of the combination that I saw was amazing.
I couldn't believe my eyes when Veselin abruptly captured the Rook. As he explained after the game, the intensive struggle tired him out, and he thought that after the Rook's capture, White would just force the draw by repetition. He did see the main idea of the combination, but he didn't think that White would just play without the Rook, exploiting King's advanced position at a4.
This move loses the game, but it also deserves an exclamation mark, because great combinations cannot take place without the partner's help. If Topalov didn't capture the Rook, the game would have probably ended in a draw, Veselin would score a half-point more, I would score a half-point less. He'd gain something, I'd lose something, and the chess world would lose a lot. But that day, Caissa was very benevolent to me. I don't know what was I awarded for, but after the d4 capture, the events were forced.
I made this move immediately. What was there to think about? The Rook is untouchable; such moves are always made with much joy. Remember me saying that the d4 pawn blocked the d-file, and there were no squares to intervene along the e-file? I was wrong. Two White Rooks open the way for other pieces into the Black camp. The construction I dreamed about - Queen at b6, Knight at a5 and Bishop at h3 - suddenly becomes true. If Black play 25... Kb8?, then after 26. Qxd4! Nd7 27. Bxd7 Bxd5 28. c4! Qxe7 29. Qb6+ Ka8 30. Qxa6+ Kb8 31. Qb6+ Ka8 32. Bc6+! Bxc6 33. Nxc6 they lose by force.
By the way, 25. Qxd4+? didn't work due to 25... Qb6 26. Re7+ Nd7, and White's attack fizzles out.
25... Kb6 26. Qxd4+ Kxa5
After the game, some players, including Anand, said that 26... Qc5 saved the game, but after 27. Qxf6+ Qd6 28. Be6!!
White block the line, getting the opportunity to make Black's position hopeless due to various threats, for instance, 28... Bxd5 (28... Rhe8 29. b4! +-) 29. b4! Ba8 30. Qxf7 Qd1+ 31. Kb2 Qxf3 32. Bf5 - the simplest: all lines are blocked, and the mating threats are unstoppable.
27. b4+ Ka4 28. Qc3
I made this move without thinking; to be honest, I couldn't make myself think because I wanted to reach the final that I already saw, and I thought (don't know why) that the game should end this way, that Black wouldn't deviate anywhere, that there's no defence. Of course, I did think for Black as well, Veselin gave me time when he was thinking, but I just couldn't get myself to search for other possibilities. Too bad. Or was it bad?
The beauty of this combination isn't worse than that of the side variant, but to be fair, I should say that from the objective chess point of view, 28. Ra7! was stronger. This move was found by Lubomir Kavalek, most probably with help from a chess computer, because it's impossible to check all the variants without an electronic partner. Still, Lubomir's idea allowed to realize all the problem-like motives in a much purer form, without giving Black any additional defensive resources that they got in the main game, even though those still weren't enough. So, after 28. Ra7 both captures at d5 lose: 28... Nxd5 29. Rxa6+!! Qxa6 30. Qb2 Nc3+ 31. Qxc3 Bd5 32. Kb2, and we get to that position where there's no defence from the Queen sacrifice at b3. Black can't get a second piece to control b3: the e6 square is attacked by the White Bishop.
The Bishop capture loses as well: 28... Bxd5 29. Qc3 Rhe8 30. Kb2 Re2. Black have defended from the Qb3+ threat by pinning the c2 pawn, but the Queen suddenly changes its route: 31. Qc7, threatening 32. Qa5#. And after 31... Qxc7 32. Ra6# the King falls victim to the Rook. An amazing mating scheme!
The best defence against 28. Ra7 is 28... Bb7, like in the main game. After 29. Rxb7 Nxd5 White have a new mating construction: 30. Bd7!, threatening to expose the King with Bxb5+ and mate it with the Rook, and after 30... Rxd7, White play 31. Qb2, threatening the b3 mate. 31... Nxb4 is the only move, and now 32. Rxd7, attacking the Queen. After 32... Qxd7, there's a mate from b4; after 32... Qc5, there's 33. Rd4, threatening to capture at b4 and then at h8; 33... Rc8 34. Qb3+ Ka5 35. axb4, and Black lose a lot of material.
The main line is 28. Ra7! Bb7 29. Rxb7 Qxd5 30. Rb6 a5 (30... Ra8 allows White to restore the material balance with 31. Qxf6 and continue the decisive attack: 31... a5 32. Bf1 Rhb8 33. Rd6, driving the Black Queen away, and the White Queen returns with mating threats). It seems that Black can defend against 31. Ra6 with 31... Ra8, but White again change the mating construction suddenly: 32. Qe3!!
Yes, to that square. 32... Rxa6 33. Kb2 (threatening to mate at b3) 33... axb4 34. axb4, threatening yet another mate at a3. The b4 capture only postpones the mate by one move: 34... Kxb4 35. Qc3+ Ka4 36. Qa3#.
The only defence is 34... Qa2+ 35. Kxa2 Kxb4+, but after 36. Kb2 Rc6 37. Bf1 (threatening to mate at a3 again) 37... Ra8 38. Qe7+ Ka5 39. Qb7, and the threat of Qb5# makes Black to give away a whole Rook.
Even worse is 28... Bxd5 due to 29. Kb2! with an unstoppable mate threat. Topalov had less than 30 minutes, I had a bit more - 32 minutes.
29. Ra7! Bb7 30. Rxb7
White pass the last opportunity to force the perpetual check with 30. Qc7. I was sure that there was more for White. Of course, 30... Rd6 31. Rb6! is a flashy, but a relatively simple variant. The Black Rook at d6 cannot perform two functions at once - defend the a6 pawn and control d4, because it's necessary for Black to meet Kb2 with Qd4. There's no check from d1, because the King just goes to a2, and the Qb3+ threat still stands. So, Black have to keep their Queen at d5 (this is a very important moment) to control b3 and be able to counter Kb2 with a d4 check. This, of course, gives an opportunity to exploit various problem-like motives that are most evident in this variant. Topalov, after thinking for several minutes, played
The most natural defence. I was counting on it, since it leads to the stunning mating finale that I've been thinking about for the last 15-20 minutes, after having a crazy vision about the final position. Nevertheless, Black had two more defences that could actually ruin this picture in my head.
1) 30... Rhe8 - this move was pointed out by Topalov the next day. This gave me a hard time in the game against Reinderman: I was so deeply engaged in calculating the variants after 30... Rhe8 (and failing to find a win) that I played the opening very sloppily, made a couple of small mistakes and flubbed everything I could. Thankfully, I was able to drive away those nightmarish visions just in time to still be able to play a good game. But Topalov's idea didn't live for long: everyone was interested in our game, and they just couldn't believe that the simple 30... Rhe8 refutes White's combination. After the next round, Ligterinck showed us a brilliant win.
White play 31. Rb6 Ra8. The seemingly obvious 32. Be6 now doesn't work: 32... Rxe6 33. Rxe6, and Black, of course, don't capture the e6 Rook since there's no defence against the mate on b3, but just play 33... Qc4. That's the counter-sacrifice I mentioned before. White are forced to capture on c4: 34. Qxc4 bxc4 35. Rxf6 Kxa3 36. Rxf7 Re8, and Black have a counter-attack and, interestingly enough, good winning chances in the endgame.
White can't allow those exchanges, and, as we see, c4 now is the critical square. With 30... Rhe8, Black could have changed the defence's construction. One Rook would protect the a6 pawn from a8, and Kb2 was now met with Qe5+, with the Rook holding e5, and Queen getting ready to come to c4.
So, the key move is 32. Bf1!!. By preventing 32... Qc4, White create a quiet threat 33. Rd6: 32... Nd7 33. Rd6 Rec8 34. Qb2; after 32... Re6, White just exchange at e6 and play Kb2; after 32... Red8, there's 33. Rc6, threatening 34. Rc5; 33... Nd7 is now met with 34. Rd6 (d-file is now blocked), and 33... Nh5 - with 34. Rc5 Rac8 35. Kb2. And again, there's no escape!
Ligterinck, presumably with computer help, found a unique defence. A counter-sacrifice that is nevertheless refuted with another computer move.
32... Re1+ 33. Qxe1 Nd7. The White Rook has nowhere to go, and, what's even more important, the Black Knight gets to b6: 34. Qc3 Nxb6 35. Kb2 this Knight gives a check from c4, and after 36. Ka2, gives another check from d2, controlling b3, and Black suddenly win. But after 33... Nd7, White have their own deflecting sacrifice: 34. Rb7!
Now, after 34... Ne5 35. Qc3 Qxf3 the simplest win is 36. Bd3 Qd5 37. Be4, and after 34... Qxb7, White play that computer move: 35. Qd1 Kxa3 36. c3, and the White Queen mates with Qc1-c2-a2. There's no defence!
I don't know if I could find all this over the board, but the beauty of this variation seems to make the combination irrefutable. Essentially, we have a problem-like mate change, which, as far as I know, never occurred in a high-level masters' game. Such an interchange isn't even characteristing for endgame studies - only for specially constructed problems.
2) Black have another counter-sacrifice: 30... Ne4! 31. fxe4 Qc4. If White just follow through with the main line after that, Black would capture at e4 with a check in the final. The White pawn would be at e4 instead of f3, making the square available for the Black Queen.
Of course, White don't have to play 32. Qxf6: after 32... Kxa3 33. Qa6 Kb4 34. Bd7 they risk nothing, but the game would most probably be drawn. 32. Qe3 also gives nothing: 32... Rc8 33. Bxc8 Rxc8 34. Qc1 Qd4! (most precise), and again, White have only a draw.
The capture at c4 leads to a complicated endgame with winning chances for White: 32. Qxc4?! bxc4 33. Kb2 f5 (best) 34. exf5 c3+ (it's important for Black to give a Zwichenschach, because after 34... Rd6 35. fxg6 c3+ 36. Ka2 hxg6 37. Bf1, mating constructions appear again: Bc4-Bb3 or Bb5-Ra7) 35. Kxc3 Kxa3 36. f6 Rd6 37. f7 Rc6+ 38. Kd4 Rxc2 39. Bf1. Perhaps the endgame is won for White due to their strong pawn and King's perspectives at g7, but White didn't begin that combination to get a better endgame. And a thorough analysis shows a better opportunity for White: 30... Ne4 31. fxe4 Qc4 32. Ra7, again creating the threat of Rxa6#.
Now, after 32... Ra8, 33. Qe3 wins: 33... Rxa7 34. Kb2; after 32... Rd1+ 33. Kb2 Qxc3+ 34. Kxc3 Rd6 there's an endgame, but different from previous examples. The Black King is still in the mating net. The pawn didn't go away from b5, and White can still pile up threats, despite the Queens' disappearance: 35. e5 Rb6 36. Kb2 Re8 (where else? 36... Rd8 37. Bd7) 37. Bg2! Rd8 (controlling d5; 37... Rxe5 38. Bb7 Re7 39. Bd5, and the Bishop gets to b3 to sacrifice itself and give a mate) 38. Bb7 Rd7 39. Bc6!!
Now after 39... Rd2, 40. Be8 wins, and after 39... Rd8 40. Bd7 Black are utterly paralyzed and have to wait until the inglorious end.
31. Qxf6 Kxa3
Here, Topalov probably still thought that White have nothing better than 32. Qxa6+ Kxb4 33. Bd7. What else can White do if their King is also threatened with checkmate? Black miss the best defence that could allow them to resist longer in the endgame: 31... Rd1+! 32. Kb2 Ra8 33. Qb6! (threatening Qa5#) 33... Qd4 (33... a5 34. Bd7) 34. Qxd4 Rxd4.
Technically, the harshest measure is probably 35. Rxf7. Black are forced to play 35... a5 36. Be6 axb4 37. Bb3+ Ka5 38. axb4+, and they can't play 38... Rxb4 due to 39. c3 - the Rook is trapped, and the endgame is technically won. And after 38... Kb6 39. Rxh7 Rc8 40. h4 White win easily: a Bishop and three pawns are much stronger than a Rook, their pieces are coordinated greatly, and they will win eventually.
But, to be honest, I didn't see that after 38... Rxb4 39. c3 the Rook is trapped, and planned to play 35. Bd7. This was also enough for win. White dominate, paralyzing the Black pieces and preparing to move their kingside pawns, since the Black Rook has to stay at a8 to prevent Bb5+ and stopping the Bishop from getting to b3. Still, they can't do it: 35. Bd7!? Rd2 36. Bc6 f5 37. Rb6 Ra7 38. Be8 Rd4 39. f4, and Black are practically stalemated; 39... Rc4 40. Bf7 Rxb4+ 41. axb4 Rxf7 42. c3 Ra7 43. Re6 a5 44. Re1, and there's a new mating construction: now Rook checkmates the King from a1.
But it was indeed a lucky day for me. Topalov captured at a3 with the King, and the variant that I dreamed about came true! I moved my fingers over the pieces and couldn't believe my eyes: my dream position would take place. Two minutes of thinking seemed like eternity.
32. Qxa6+ Kxb4 33. c3+! Kxc3 34. Qa1+ Kd2
There's no way back: 34... Kb4 35. Qb2+ Ka5 36. Qa3+ Qa4 37. Ra7+, winning the Queen.
35. Qb2+ Kd1
35... Ke3 36. Re7+ Kf3 37. Qg2# - another mating finale. The Black King came to its Golgotha all the way from b8 to d1!
And now, when it seemingly reached a save haven (White have no checks anymore), the h3 Bishop that was tasked with guarding e6 makes its strike.
White have attacked the Queen, and it has nowhere to go: if it goes along the c-file, there's 37. Qe2#, and 36... Qe6 is met with 37. Qc1#. The Bishop is untouchable as well: 36... Qxf1 37. Qc2+ Ke1 38. Re7+. Hardly anyone would like such a simple mate, especially compared with earlier constructions.
Black mount a counter-strike, and for a second, it seems that the worst is over: what reserves do White have? Black are getting ready for the checkmate, but now the White Rook jumps into action.
The a1-h8 diagonal weakness is a very important element of the combination. Everything usually hinges on such small things. If the Black Rook stood at g8, there would have been no combination... But now, after Rd7, it's hopeless for Black. Though Topalov still tried to resist. Black are forced to capture the d7 Rook.
37... Rxd7 38. Bxc4 bxc4 39. Qxh8 Rd3
An illusion of activity: if White capture at h7, then after 40... c3, the pawn gets to be queened. But that's no checkers: capture isn't necessary, and now the Queen shows its full strength.
40. Qa8 c3 41. Qa4+ Ke1 42. f4
Depriving Black of their last hopes. White have too many pawns for Black to hope to get a Queen vs. Rook endgame.
42... f5 43. Kc1 Rd2 44. Qa7
The Queen starts to attack the Black pawns, and the h2 pawn is untouchable due to Qg1+. That's how this amazing game ended.