What do masters do when their opponent deviates from opening book or they enter a position where they do not know the book moves?

I want to know how they think,the guidelines of choosing the next move.

They use their understanding of the position, pawn structures etc.

It's not like they have only memorised the book moves, they know what they are trying to achieve with them.

they go cry in a corner

No seriously, they've problably seen everything. I read somewhere that GMs have literally thousands of positions memorized.

There's no way you're gonna play something that a GM didn't see already, a bit scary thought.

Scottrf is right, they understand why the book moves are played and why the non-book moves aren't considered best.  So when someone leaves book or they forget the moves, they use their knowledge of chess.

I'm not too sure about that.

If he plays 1 e4 and I reply 1... b5, I doubt he thinks "oh, yeah, this is what I play against that..."

Thousands of positions? Absolutely. That said... that's not really all that many positions in terms of how many are possible.

Korchnoi and Petrosyan have already said that only UNDERSTANDING a position is the hallmark of Great Masters!

So, They use their positional lessons which they have received in young age or innate Positional Genius like Capablanka!

raykrish, at your rating, this book may help you, as it did me:

LOL, at his peril, I have a few wins with that

But seriously, at our club there was a Sokolsky line that was fashionable (Yeah, I know, the club is weird).

A few years ago, GM Neil McDonald visited (I wasn't there, so this is hearsay) and the players told me he was totally unfamiliar with this line, and a little intrigued, as I recall someone managed a draw in a simul, or a Blitz afterwards, or something.

Since people at that level have a deep understanding of chess they'll figure out good moves over the board. It was said that Capablanca once refuted Marshall's home preparation over the board, so they do it by ability, not memorization.  If we don't understand something in one of our openings we should strive to understand it by weighing different moves against the principles of chess. 

I think 1000 or 2000 positions will cover pretty much everything that there in chess (or at least everything that can happen with one's repertoire).

Don't be misled by the game tree complexity of chess. While the possible permutations of pieces are bazillions, the actual positions that could happen in practice are much, much less. Also in the endgames there are many permutations of pieces that would not alter the evaluation.

For example: Lucena is one position. If you count all the possible locations of the black and white rooks and the black king, then multiply all that for the 6 files, you have more than 1 million...but the knowledge required for that position is just one.

ok, if they don't know what to play, they will just sit their and think until a good move pops into their heads

I saw this thing on 60 minutes about Carlsen that said he usually already knew what to do the moment his opponent made his/her move and he just made sure that it was sound before making that move.

raykrish you ask an important question.  That phase of the game would be associated with the term Theoretical Novelty.  The top players spend a huge amount of time over the board when a Theoretical Novelty appears...

Grandmasters aren't mere "memorizers". They really understand chess in depth...

i think when a gm meets a non book move ...hmmm...they go take a leak

and return ith a great move similar to the best move of a chess program or somthing.

@pfren, what I think is most interesting about your game, is not when it leaves "opening book" but when it enters "endgame book", specifically for instance when it essentially becomes a position out of Minev's "Practical Guide to Rook Endings"