Which side should a beginner castle?

Kingside, nearly always. It needs two squares cleared instead of three, so it happens three or four moves earlier, and the king ends up in a corner behind three pawns that have not moved. Long castling is a good move in the right structure, but it needs you to know what you are doing with the g and h pawns afterwards.

What is opposite-side castling?

When one player castles short and the other castles long, so the kings sit on opposite wings. It usually turns into a race: each side throws the pawns in front of the enemy king forward, because those pawns are no longer sheltering anybody. Games like this are sharp and short, and one tempo often decides them.

Can I castle on the side my opponent castled?

Of course, and it is the quieter choice. With both kings on g1 and g8 neither player can push the g and h pawns without exposing their own king, so play tends to move to the centre and the queenside instead. If you dislike sharp attacking positions, castling the same side as your opponent is the practical way to avoid them.

Kingside vs queenside castling

Two squares to clear versus three. That is the first difference, and it decides most games at club level before anything about strategy comes into it. Short castling needs only f1 and g1 empty, so a knight to f3 and a bishop to c4 or e2 and you are done by move four. Long castling needs b1, c1 and d1 empty, which means the knight, the bishop and the queen all have to find homes first.

Safety against activity

After O-O the king sits on g1 in the corner, with f2, g2 and h2 untouched in front of it and the rook next door on f1. It is the tidiest arrangement in chess: no gaps, nothing hanging, and the h2 pawn defended by the king itself.

After O-O-O the king stands on c1, one file further from the corner, with a hole on b1 and the pawn on a2 undefended. In exchange the rook lands on d1 immediately, on the file where the action usually is, without spending a move getting there. A short castler often needs Re1 or Rd1 afterwards; a long castler already has the rook where it wants to be.

That is the trade in one sentence. Short buys safety, long buys a tempo and a better rook.

Opposite wings

The interesting case is when the two of you go different ways.

White on c1, Black on g8. The g-pawn is already on g4 and h4 comes next; Black has a6 and b5 rolling and wants b5-b4 to shift the knight on c3.

Once the kings are on opposite wings, the pawns in front of your own king are guarding you and the pawns in front of theirs are ammunition. So both sides push. White throws g4, h4, g5 at the black king; Black answers with a6, b5, b4 at the white king. Nobody has to worry about weakening their own shelter, because the attacking pawns are on the far side of the board.

These games are decided by one tempo more often than by one pawn. If you are two moves faster than your opponent, you can afford to give up a piece to get the lines open first. If you are two moves slower, that same sacrifice is a resignation. The way to feel the speed is to solve positions of both kinds: kingside attacks against a castled king on g8, and queenside attacks against a king on c1.

How I choose

Three questions, in this order.

Where are my opponent’s pawns pointing? If Black has played g6 and Bg7, the kingside is solid and hard to attack, so my rooks and my king often go the other way.

Which centre file is going to open? The rook I most want on that file should be the one that does not have to travel.

And how fast do I need to be? If I am behind in development, I castle short and get on with it. Long castling with a loose position invites everything to arrive on c2 and b2 at once.

If none of that is clear yet, castle short. It is the move you will play in most of your games, and it is almost never wrong. There is more on how the two set-ups feed into whole opening plans in castling queenside.