Shereshevsky’s “Associative Thinking” on 5 versus 4 rook endings with passed a- and b-pawns

Part 1

Mikhail Shereshevsky is widely recognized as one of the top chess trainers in the world, renowned for his clear, methodical approach to teaching. He places special emphasis on the study of the endgame, which he considers essential for developing a deep and practical understanding of chess. His works, particularly those devoted to endgame strategy, have become classics and remain highly influential across generations of players.

In his recent book Associative Thinking: How to Connect Patterns and Creativity in Chess (New in Chess, 2025), a translation of the Russian original first published in 2024, Shereshevsky examines the role of associative thinking in rook endings (Chapter 7). His explanations are methodically instructive, and many of the positions he analyzes involve endings with a passed a- or b-pawn. The quality of the analysis is generally high, though there are a few errors. 

D. 1



According to Shereshevsky, this position should be evaluated as drawn. However, a closer examination reveals that his assessment is mistaken: the position is in fact winning for White.

1.Ke3! Kf6 

White has two equally effective winning plans: advancing the a7-pawn or bringing the king to support the passed pawn. 

D. 1-1



In this position, Shereshevsky surprisingly limits his analysis to 2.Kd3?, which throws away the win. This premature concession of the f2-pawn fails because the White king is still too far from effectively supporting his passed pawn. 2...Rxf2! 3.Rc7 Ra2 4.a7 Ke6 5.Kc4 f5! 6.Kb5 fxe4! 7.Rc6+ Kd5! 8.Ra6 Rb2+! 9.Ka5 Kc5! 10.Rc6+ Kd4!=.

A) 2.Ra8  

Or 2.f3 (B).

D. 1-2



1) 2...Ra3+ 3.Kd2 Ra2+ 4.Kc3 Rxf2 5.Kb4 Ra2 6.Kb5 Ke7 7.Kb6 Rb2+ 8.Ka7+-.

If Black declines to capture the f2-pawn, White wins without much difficulty, as demonstrated in the analysis of the Spassky–Torre (1982) endgame after 42...g6 (https://www.chesspub.com/cgi-bin/chess/YaBB.pl?num=1719207215). It is surprising that Shereshevsky, having already shown the winning a7 plan in the variation proposed by Shoshin (Diagram 2), does not attempt to apply the same method in this position, which is even more favorable for White.

2) 2...Ke7 

D. 1-3



3.a7 Ra3+ 4.Kd2 Ra2+ 5.Kc3 Ra3+ 6.Kb4 Ra1 7.f3 Ra2 8.g4 Kf6 9.g5+ Ke7 

D. 1-4



9...Kg7 10.Kc5 Ra6 11.Kd5 Ra5+ transposes to D. 1-6.

10.Kb3

White can only win by sacrificing f3-f4. Therefore, he should play the king down to neutralize the black passed pawn. Compare with D. 4.

10…Ra1 16.Kb2 Ra4 17.Kb1 Ra5

17...Rb4+ 18.Kc2 Ra4 19.f4+–․ 

18.f4+–․

3) 2...Kg7 

D. 1-5



3.a7 Ra3+ 4.Kd2 Ra1 5.f3 Ra3 6.Kc2 Kf6 7.Kb2 Ra6 8.g4 Kg7 9.g5 Rb6+ 10.Kc3 Rc6+ 11.Kb4 Ra6 12.Kb5 Ra3 13.Kc6 Ra1. 

See the analysis of Spassky-Torre, line 42...g6, 47...Ra1 48.g3 ...66.Kc6 Ra3 67.Kd5. 

D. 1-6



12.Kd5 Ra5+ 13.Kd6 Kh7 14.Kc6 Ra1 15.Rd8 Rxa7 16.Rd7+–.

Back to D. 1-1

B) 2.f3 

D. 1-7



2...Ra3+ 3.Ke2 Ke6 

Ra2+ 4.Kd3 Ra3+ 5.Kc4 Rxf3 6.Rb7 Rxg3 7.Rb3 Rg4 8.a7 Rxe4+ 9.Kb5!+–. 

D. 1-8



Now White can directly implement the a6-a7 plan with 4.Ra8. 4.g4 seems risky but it also wins. 

4.g4 hxg4 5.fxg4 Ra4 

D․ 1-9



It looks like Black is securing a draw by attacking the e4-pawn, but it turns out that White can ignore this threat.

6.Kd3 Rd4+ 7.Kc3 Rxe4   

8...Ra4 9.h5! gxh5 10.gxh5! Kf6 11.Kb3! Ra1 12.Rg8+– (12.Kb2 Ra5 13.Rg8!+–)

9.g5+! Kg7

After 9...Kf5 White can finish the game with a single tactical stroke. 10.h5! Kxg5 11.h6 Kxh6 12.Rh8+ Kg5 13.a7 Ra4 14.a8Q Rxa8 15.Rxa8+–.

10.Kb3! Re1 11.Kb4 Rb1+ 12.Kc5 Rc1+ 13.Kb6 Rb1+ 14.Ka7 f5 15.gxf6+! Kxf6 16.Rb8 Ra1 17.Kb7 e4 18.a7! g5 19.h5+– or 19.Rf8+ Kg6 20.Rg8+!+–.

Shereshevsky reports that his student, FIDE Master K. Shoshin, recently discovered a new winning method by isolating and attacking Black’s f7-pawn in the following position:

D. 2



The main line of the analysis of K. Shoshin is as follows: 

1.a7 (with !! mark by Shereshevsky) Ra3 2.f3 Ra2+ 3.Kf1 (see D. 3-1) Kh7 4.Ke1 Kg7 5.Kd1 Kh7 6.Kc1 Kg7 7.Kb1 Ra3 8.Kb2 Ra5 9.Kb3 Ra1 10.Kc4 Ra5 11.Kb4 Ra1 12.Kc5 (see D. 3-2) Kh7 13.Kd5 Ra5+ 14.Kd6 Kg7 15.g4 (see D. 3-3) Kf6 16.g5+ Kg7 (see D. 3-4) 17.Kc6 Ra1 18.Kd5 Ra5+ 19.Kd6 (see D. 3-5) Kh7 20.Kc6 Ra1 21.Rd8 Rxa7 22.Rd7 Ra1 23.Rxf7+ Kg8 24.Rf6.

This analysis is entirely correct. However, the idea and its analysis are not new, as I had already examined all the essential variations after 1.a7 Ra3 2.f3 Ra2+ 3.Kf1 in detail on this forum, in the thread Spassky–Torre.

D. 3 
(pgn-file Spassky-Torre, Line after 42…g6 43.Ra7! Kg7 44.a6 Kf8 45.Ra8+ Kg7 46.Kh2! Ra1 47.g3 Kf6)




48.a7 Ra2+ 49.Kg1 Kg7 50.Kf1

D. 3-1



50...Kh7 51.Ke1 Kg7 52.Kd1 Kh7 53.Kc1 Ra3 54.Kc2 Kg7 55.Kb2 Ra6 56.Kb3 Kh7 57.Kb4 Ra1 58.Kb5 Kg7 59.Kc5

D. 3-2



59…Ra2 60.Kd6 Ra5 61.g4 


D. 3-3



61...Kf6 62.g5+ Kg7 


D. 3-4



63.Kc6 Ra1 64.Kd5 Ra5+ 65.Kd6+

D. 3-5



62...Kh7 63.Kc6 Ra1 64.Rd8 Rxa7 65.Rd7! Ra6+ 66.Kb5 Ra1 67.Rxf7+! Kg8 68.Rf6!+-

In the Shereshevsky and Shoshin analysis of the position from Diagram 2, the line 12…Kf6 (instead of 12…Kh7) 13.g4 Ra2 14.g5+ Ke7 is especially noteworthy. This variation was not included in my analysis of Spassky–Torre.

D. 4



15.Kb4 Ra1 16.Kc3 Ra3+ 17.Kb2 Ra4 18.Kb1 Ra3 19.f4 exf4 20.Kc1 f3 21.Kd2 f2 22.Ke2 Ra2+ 23.Kf1!+-.

In D. 3 1.a7 is not the only winning move. White can also win with 1.f3 or 2.Ra7.   

D. 5



In the Russian edition, in his analysis of the game Spassky–Torre (1982), Shereshevsky evaluated the position after White’s 42nd move as drawn: “In this drawn position the Filipino grandmaster made a gross logical error and played 42...f6??”
However, I have shown on this forum that Black’s position is already lost: Black loses not only after Torre’s move 42...f6, but also after Shereshevsky’s recommended 42...g6. It is good to see that the English edition has reached the same conclusion, presenting only the position after 42...f6 had been played.