A Rook Endgame

An interesting rook endgame from a recent game.  How do you assess this position?  Is White winning, or can Black force a draw?  How should each side proceed?

Both my opponent and I came to the wrong conclusion.  I'll post the answer later.

first Rd7 to get the king out the way. if K b7 then Kc6!. then move the pawn

kco pointed out Rd7 hangs the rook. Also, Kb7 would be illegal, and if it was legal, Kc6 would be an illegal response to it. Maybe you meant Kb8 instead of Kc7, but then Rd8# is better than Kc6.

The best plan I've been able to thin of for white is 1. Rf8+ Kd7 2. Ra8 followed by Ra5, Ka7 and walk the pawn up the board, using the pawn to shield from any checks from the side. I haven't worked out if this actually works.

What's the plan for white after 1. Rf5. Let's say black just does the simplest thing and keeps his rook on the b-file, 1. ... Rb1.

I think it would have to be Ra4, not Ra5 otherwise there's nowhere to run from a check along the fifth rank but could black not just pin the pawn to the King in that plan anyway?  The pawn can't advance, and every time the King moves to a6 Black could just kick it back with Rd6+ (or similar, depending on the Rook's file).

It's like pulling teeth. What is your plan, not just your next move.

1. Rf5 Rb1

2. Kc6 Rc1+

and now the king has to go back where it came from because Rc5 Rxc5+ Kxc5 Kb7 is a draw.

Grobe, I think the problem with my plan is that the black king can get back over while the rook is on a5. So my new plan is to put the rook on b8. Then white can play Ka7 (rook protects pawn) and either push the pawn if not checked, or if Ra2+ play Kb7 and then push the pawn. Now the white king keeps the black king out of c8 and the white rook can go around to a6.

A table-base will be flawlessly reliable by definition.

If I remember correctly 2. Rb8 as suggested by Loomis is a standard idea to win this type of ending - might even be the only way to make progress but I'm not sure about that. 6. Ra8 + Ra6 apparently wins but I think it would also be possible to methodically build Lucena (http://en.wikipedia.org/wiki/Lucena_position) type position as follows:

Loomis has it right, it's a win for White by playing Rf8+ followed by Rb8.  Nothing else really works:

• Rf8+ Kd7 Ka6 Kc7
• Rf5 Rb1 Ka6 Ra1+ Kb6 Rb1
• Rf5 Rb1 Rc5 Kb8 (this was the game continuation)

I never found the winning plan in my analysis, and neither did my opponent.  It wasn't until I analyzed with a computer afterwards that I found it.

kco. I hate to say it, but this is another piece of information that doesn't actually help me win the position. Great, there is a win in 31 moves. How do I do it? (don't need to answer, other posters have shown good ideas).

You need to show the line, otherwise this just looks like speculation.