RPP vs R (two connected pawns) was not enough to win!

[Ajedrecista]

Hello:

I played other interesting-to-me 5+0 game today:

[pgn][Event ""]
[Site ""]
[Date "2026.04.15"]
[Round ""]
[TimeControl "300+0"]
[Result "0-1"]

1. d4 d5 2. c4 e6 3. Nc3 h6 4. Nf3 Nf6 5. cxd5 exd5 6. e3 c6 7. Be2 Bd6 8. O-O Bf5 9. Bd3 Bxd3 10. Qxd3 Nbd7 11. a3 O-O 12. Bd2 Re8 13. Rac1 a6 14. Na2 Ne4 15. Bb4 Ndf6 16. Bxd6 Qxd6 17. Nc3 Nxc3 18. Rxc3 Ne4 19. Rcc1 Rab8 20. Nd2 Qe6 21. Nxe4 Qxe4 22. Qxe4 Rxe4 23. Rc3 Rbe8 24. Rb3 R8e7 25. Rc1 f5 26. Kf1 f4 27. Rcc3 fxe3 28. fxe3 g5 29. Ke2 Kg7 30. Rb6 Rxd4 31. Rcb3 Rde4 32. Rxb7 Kf6 33. Rxe7 Kxe7 34. Rb6 Kd6 35. Rxa6 g4 36. b4 Kc7 37. Ra8 Kb7 38. Rh8 Re6 39. Kd3 Kc7 40. Rg8 h5 41. Rg5 Rh6 42. e4 dxe4+ 43. Kxe4 Kd6 44. Ra5 Re6+ 45. Kd3 Rf6 46. Ra6 Kc7 47. b5 Kb7 48. Rxc6 Rf2 49. Rc2 Rf5 50. a4 Kb6 51. Ke4 Rg5 52. Kf4 Rd5 53. Kg3 Ka5 54. Ra2 Rd4 55. h3 gxh3 56. Kxh3 h4 57. Ra3 Rb4 58. g3 hxg3 59. Kxg3 Rd4 60. Kf3 Kb6 61. Ke3 Rh4 62. Kd2 Kc5 63. Kc2 Kb4 64. Ra1 Rh2+ 65. Kd3 Rh4 66. b6 Rh8 67. a5 Rb8 0-1[/pgn]

I lost on time while I had winning chances most of the time when two pawns up, before oversimplifying. Black's last move was a blunder and I almost played the winning move 68.- a6, but I flagged instead.

Nevertheless, I was surprised that two connected pawns were not enough to win in a RPP vs R endgame. I know that there is a (difficult) technique to draw this endgame against f&h pawns (alternatively, a&c). Here is the first position with six pieces OTB:

[d]8/8/8/kP6/Pr6/R5K1/8/8 b - - 0 59

Syzygy online EGTB site says that 'KRPPvKR statistics (unique positions)' are only 76.0% of wins for the pawnful side, but I guess that most of the non-winning positions are immediate captures and some cases with a&c or f&h pawns, as said before.

The endgame is easy for engines, but I wanted to share it, anyway. A more related question to computer chess would be to compute or estimate the percentage of RPP vs R endgames with connected pawns and without immediate captures that are not winning for the pawnful side.

Regards from Spain.

Ajedrecista.

[chrisw]

Great KR vs KRPP by Paul Keres with a and c (or f and h) pawns, how to draw, useful for swindling, I recollect from my cafe chess days in 1970 or so.
Checked with prompt to Gemini "Paul Keres analysis of KRvKRPP where the pawns are on the f and h files" which even brought it up, with the book reference.

[IanKennedy]

These positions are dealt with in the Muller-Konoval book Understanding Rook Endgames, all verified with EGTBs. A brief summary is
1. Two connected passed pawns are usually winning
2, The main exceptions arise when the defending king manages to get into the 'hole' of the pawns..[as in your diagram]
2a. But not all such positions are drawn...

pp25-27

[chrisw]

These positions are dealt with in the Muller-Konoval book Understanding Rook Endgames, all verified with EGTBs. A brief summary is
1. Two connected passed pawns are usually winning
2, The main exceptions arise when the defending king manages to get into the 'hole' of the pawns..[as in your diagram]
2a. But not all such positions are drawn...

pp25-27

All well and good, but the above book is from computer chess era using EGTB. The fundamental human theory (on a,c and f,h) came from the Soviet Union way earlier, of which Keres is deemed gold standard.