This should be a draw. Black has an extra pawn, but White has the possibility of exchanging the c5-pawn for the b7-pawn in order to reach a drawn ending of R+p vs R.

81. Rd1? This is wrong and should lose. After 81. Rc1 Black would not be able to make progress, since 81... Rd4+ 82. Kc7 Rc4 83. Rh1 Kxc5 84. Kxb7 only leads to a draw. 81... Rb3? Missing the winning line 81... b6! 82. cxb6 Kxb6 In endings of R+p-R, it is much worse for the defender to have his king cut off along a rank than along a file, and here there is no way to prevent the advance of Black's pawn, for example: 83. Rc1 (or 83. Rd2 c5 84. Kd5 Kb5 85. Rd1 Rb3 and so on) 83... Rd4+ 84. Ke5 c5 85. Rb1+ Kc6 86. Rh1 or else ...Rh4 followed by ...c4 86... Rd8 87. Rc1 Re8+ and Black wins. 82. Rc1 Rh3 The only try for a win, even though it allows an exchange of pawns. 83. Rb1+ Kc4 84. Rxb7 Rh6+

85. Ke5?? 85. Kc7 Kxc5 86. Rb1 (The full checking distance is necessary: 86. Rb2?? Rh7+ 87. Kc8 Kd4 wins) 86... Rh7+ 87. Kc8 It is surprising that Black cannot win even though White's king is cut off on the back rank, but after 87... Kd5 88. Rd1+ Ke4 89. Rc1 it becomes clear that there is nothing Black can do. 85... Kxc5 Now White's king gets cut off and there is no hope of saving the game. 86. Ke4 Kc4 86... Rh3 also wins. 87. Ke3 Kc3 88. Rc7 Re6+ 89. Kf4 Kc4 90. Kf5 Rh6 91. Ke4 c5 92. Ke3 Rd6 93. Ke2 Kb4 94. Rb7+ Kc3 95. Rb1 Kc2 96. Ra1 Re6+ 97. Kf3 c4 White resigned. [John Nunn]