1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 g6
This
is the move that defines the Dragon Sicilian. This opening is named
after the pawn structure h7–g6–f7–e7–d6 that has the shape of
a Dragon and, incidentally, the name also reflects the spirit of the
variation. The Dragon is notoriously ambitious, risky and reckless.
The Dragon has always been a popular and very successful opening for
amateurs and club players, but it can be played all the way up to the
very highest level. An important characteristic of this setup is that
we have basically no weaknesses (the e7–pawn at the base of the
chain barely counts), but it does create a target for a kingside
attack and White’s most principled approach of castling long does
ask Black some important questions. The Yugoslav attack lines, in
which White castles long, are quite important to know move by move,
while slower setups with the short castle can be played ‘by ear’.
Though to get that ‘ear’ tuned right, it is still good to go over
all the variations in great detail and study some model games, but
when it comes to memorizing a lot of concrete lines, the long castle
beats the short castle hands down. One good thing about the Dragon is
that, if White does not take up the challenge and go for the long
castle plans, then he should abandon all hope for an opening
advantage and Black gets easy and comfortable play thanks to an
excellent pawn structure, harmonious development and the semi-open
c-file.
6.Be3
By
far the most played and also the most dangerous setup, called the
Yugoslav Attack. True to the nature of the position, White wants to
quickly castle long and start an attack on the kingside. You will
encounter this in most of your games, no matter what your level is.
Both the amateurs and the professionals are in sync here and most of
them take up the challenge and aim for the long castle plans. And
rightly so. The Yugoslav Attack is the core of your Dragon Sicilian
repertoire.
6...Bg7
The
idea of the Dragon is to develop the bishop to the long diagonal,
while also keeping the central pawn structure e7–d6 intact.
Interestingly, this often means that, besides going for an attack on
the queenside, Black is also often happy to enter an endgame, as his
pawn structure is excellent and there is not much White can offer in
return for our pressure along the semi-open c-file. White, on the
other hand, will try to trade off the pride of our position (the
g7–bishop) with Qd2 and Bh6 (or Bd4 in some cases) and use the
g6–pawn as a target for an attack on the kingside with h4–h5.
7.f3
White
needs to secure the bishop, as 7.Qd2 can be met with 7…Ng4!
7.Be2
White is aiming for short castle and throws away all his chances of
an initiative.
7...0–0 8.0–0 Nc6 9.f3 d5 10.exd5 Nxd5 11.Nxd5
Qxd5 12.Nxc6 Qxc6
Black has an edge.
7...0–0
7...Nc6
Better castle first!;
7...a6
This
is called Dragodorf as it combines both the Dragon and Najdorf. Black
is adviced not to play like this, as White often prepares to castle
long and often gets the better chances.
8.Qd2
8.Bc4
A very good alternative!
8...Nc6
A)
9.Bb3
This
is quite often played, but there is a clever way for Black to get
into familiar lines of play with 9…Qa5! 9...Qa5! 10.Qe2? This
innocent move is simply wrong. (10.Qd3
This
is a very odd move here, now that we still have the knight on c6,
ready to jump to e5 with a tempo. Engines like this move. It is
interesting, says in Anish Giri in his Chessable course. 10...Bd7
Also
very logical is to get Nd7–Nc5 in first. 11.0–0–0
Rfc8 This
happened in one correspondence game. 12.Kb1
Rab8 Black
keeps the option of Ne5. 13.h4
Ne5 14.Qd2
White
lost a tempo on Qd3–Qd2, which can’t be great in an opening as
sharp as the Dragon. Still, the position is very complex. 14...b5
15.g4 White
pushes his kingside pawns further in typical fashion of the Dragon.
15...Nc4 A
typical way for Black of getting good play along the soon opening
b-file. 16.Bxc4
bxc4 17.Ka1 White
has to take care of his queenside, with the open b-and a-files.
17...h5
18.gxh5 Nxh5 19.Rhg1 Rc5 20.a3 Qa6 21.Rg5 Ra5 22.Rxa5 Qxa5 23.Nde2
Bc6 24.Bd4 Bf8 25.Qg5 Qxg5 26.hxg5 a6 27.Kb1 e6 28.Kc1 ½–½
(28) Moreto Quintana,A (2448)-Tseng,W (2490) ICCF email 2018)
10...Nxe4!!
White loses material.
11.Nxc6 Bxc3+
Black
is winning.;
B)
9.0–0 Qb6
Black
has at least equalised.
10.Bb3
Ng4! 11.fxg4 Bxd4 12.Bxd4 Qxd4+ 13.Qxd4 Nxd4 14.Nd5 Nc6 15.h3 Kg7
16.Rf2 h6 17.Rc1 Bd7 18.c4 Rad8 19.Rcf1 Be6 20.Nf4 Rc8 21.Nxe6+ fxe6
22.Rxf8 Rxf8 23.Rxf8 Kxf8 24.Kf2 Na5 25.Ke3 e5 0–1 (40) Saharinen,P
(1901)-Serna,J (1960) ICCF 2024]
8...Nc6
This
is the main tabia of the Dragon Yugoslav Attack. Here, White has to
choose between the old main line with 9.Bc4 and the more modern
approach with 9.O-O-O.
9.Bc4
This
is the old main line and a move you are likely to encounter a lot as
a Dragon player.
9.0–0–0
This
is the modern main line and a very critical continuation. The
established theory is that Black is defending a slightly worse
position here, making the whole Dragon a lot less attractive than
what it used to be. However, White needs to now exactly how to
proceed.
9...d5
Theory
implies that whenever we get the chance to execute the ...d5 break we
should take it.
10.exd5
This
is the main line, although there are quite a few tricky sidelines
that I do not cover in this short summary.
10...Nxd5
11.Nxc6 bxc6 12.Bd4
This
is possibly the main line at the highest level these days and Anish
Giri can understand why, as the trend has shifted from the ambitious
12…e5 towards the defensive 12…Bxd4 . After the trade of bishops,
White gets a strategically superior position due to his better pawn
structure and also a safe king. Instead of going for the defensive
approach, GM Giri proposes in his great Chessable course to look for
a solution with the old master (who have kept the Dragon spirit alive
by preserving the dark-squared bishops) and inject the old main line
with some new knowledge from AI engines. They can save practically
anything nowadays.
12...e5
Played
in order to keep the g7–bishop alive for now, instead of the modern
solution with 12…Bxd4 13.Qxd4 Qb6, where Black is hoping to hold a
solid, but not a very fun position. With the help of the most
powerful modern chess engines, Anish Giri was able to prove that the
old masters weren’t betrayed by their intuition when they decided
to go for the natural 12… e5 here. The good old Dragon players have
always felt that this was the way to go in the Dragon, but the theory
and engines have always come in the way. Now, with the newer and
stronger neural network-based engines, Giri was able to make this old
move work. Keeping the bishops gives Black something to hope for. It
also is still the most popular move. Let’s keep it this way.
13.Bc5
Black
should not yet worry about Bxf8 as he would get fine compensation in
dark-square play. (13.Ne4
exd4 0–1
(13) Rivas Romero,G (2145)-Taylor,B (2128) ICCF 2024)
13...Re8
This
is a case of the simpler, the better. The sophisticated main
theoretical move 13…Be6 is flashy and clever, as taking the
exchange on f8 would be a bad idea for White. The problem is that
White doesn’t usually take the exchange.
14.Ne4
This
is the most challenging move.
14...Qc7
The
point. We don’t want to spend time on Be6 just yet, explains Anish
Giri. With the very rare 14…Qc7, Black prepares the crucial
f5–push, which now becomes possible, as Nd6 can be met by Rd8. The
queen frees up the d8–square for the rook.
15.h4
The main move. White assumes that f5 is not really a threat and
starts direct kingside play. He assumes wrong.
15...f5
Now
is the moment! Things seem to work out for Black in the end!
16.Nd6
Rd8 17.Nxc8 Raxc8
Black
has a ‘working’ pawn majority in the kingside-center area and a
potentially strong bishop on g7. The complications seem to work out
and the engines eventually give in even further and admit that Black
has nothing to fear here. Black has achieved a beautiful pawn
structure on the kingside and White is yet to start his play. This is
punch in the face for those that say that the Dragon Sicilian is
analyzed to the end. White will be forced to finding the best
possible move in very complicated positions. Not an easy task!
18.h5
Nb6
This
all happened in a rather anonymous correspondent game Holec-Eschert,
2011. Black held it and it is actually hugely important for the
theory of the Dragon, says Giri. Several modern correspondence chess
games have studied this line in pretty much detail. Beautiful draws
have surfaced!
19.Bxb6
Going
for opposite colored bishops is a safe practical attempt by White.
The endgame is a draw, even though White ends up on the pressing
side.
19...Rxd2
20.Bc4+
A
strong intermezzo, connecting the rooks.
20...Kf8
21.Bxc7 Rxd1+ 22.Rxd1 Rxc7 23.hxg6 Bh6+ 24.Kb1 hxg6 25.Rd6
The
only try to keep the game going, otherwise, Ke7 would stop the rook
from entering.
25...Kg7
After
this move, let me just point out some correspondence chess games.
A)
26.b4
Bf4 27.Re6 Rb7 28.a3 a5 29.Kb2! ½–½ (53) Sauceda Salazar,E
(2497)-Weglicki,S (2375) ICCF 2025 (29.c3
½–½
(53) Szerlak,A (2400)-Kazoks,A (2367) ICCF 2024)
;
B)
26.Re6 Rd7 27.b4 Be3 28.Rxc6 Rd2 ½–½ (39) Desquiens,S
(2111)-Sonnet,J (2190) ICCF 2024]
9...Nxd4
Anish
Giri's main recommendation. It is certainly possible and highly
interesting to investigate the slower lines starting with Bd7 and
then Rc8 or Rb8. But while some of that is certainly a lot of fun,
Black will experience some theoretical difficulties if White knows
what he is doing, and it is certainly one of the reasons why the
Dragon has faded away from high-level practice.
10.Bxd4
Instead,
10.Qxd4 would be a big mistake due to 10…Ng4 11.Qd2 Nxe3 12.Qxe3
Bd7 , and with the bishop pair and the semi-open c-file, Black has
all the fun.
10...Be6
It
is amazing how well the bishop is placed on e6, fighting for some
vital squares along the a2–g8 diagonal. The idea is to develop the
bishop to e6 in one go, now that the white knight is no longer on d4.
This is a rather old line that has always been there, but was not
thought to give Black adequate play. Now, with the help of modern
computers, Anish Giri has managed to fully rehabilitate it and the
old conclusions were completely overthrown. Besides being the top
choice of the best engines against the 9. Bc4–system, this approach
is also very logical and very direct. We want to develop our pieces
to the most active squares right away and start our play on the
queenside as soon as possible.
11.Bxe6
The
bishop trade does double Black’s pawns, but it strongly reinforces
the central control, as the important d5–square will now be guarded
by an e6–pawn.
11.Bb3
Apart
from this being the most popular move, it also feels right. White
parks the bishop on b3 and asks Black what he has achieved by playing
Be6.
11...Qa5
The
perfect square for the queen, from where it not only applies pressure
on the c3–knight and the a2–pawn, but also very importantly
prepares the b5–b4 push. The standard way of developing the
initiative in the Dragon.
12.0–0–0
transposes
to a popular line of play.
12...b5
There
is a ton of old theory with 12…Rfc8 first too, but the modern
computers are in agreement that the direct pawn push is the way to
go.
13.Kb1
This
is an important move whenever there is an x-ray between the queens on
d2 and a5. Now Black has to watch out for the Nd5 Qxd2 Nxe7+
intermezzo. This is a very important motive in the Dragon.
13...b4
Forcing
the matters. Continuing with the direct approach.
14.Nd5
The
most principled and most common reply.
14...Bxd5
15.exd5
Michail
Tal's improvement.
15...Qb5
Pressing
down the b3–bishop and intending a5–a4. White will end up having
to trade queens with Qd3 or Rhe1/Qe2 and the ensuing endgame is
completely fine for Black.
16.Qd3
Trading
queens does save the b3–bishop.
16...Qxd3
17.cxd3
This
can either be met by 17…Nd7 which brought Zvjaginsev victory over
Pershin in the 2018 WCh rapid, or by 17…a5 which after 18.Rhe1
transposes to 16. Rhe1.
17...a5
18.Rhe1 a4
With
a direct transposition to the 16.Rhe1 variation. Grabbing some space
first.
19.Bc4
Now
it’s time to address the e7–pawn. Or is it? asks Giri.
19...Rab8
This
is the easiest, improving on Moranda-Klekowski, 2018. We take control
over the b5–square to prepare Rfe8, as it turns out that the
e7–pawn is not really hanging.
20.Rxe7
Nxd5 21.Bxd5 Bxd4
With
equality. The opposite-colored bishops are famously neutralizing each
other and, with Rb5 incoming, White will not be able to keep up the
pressure on the f7–pawn. But first, we would want to go a3 to
create a far-advanced pawn there, just in case our rook gets
activated one day.)
11...fxe6
12.0–0–0
White
has to castle long if he wants to get anything in this particular
line, but also in the Dragon in general.
12...Qa5
The
queen belongs on this square, connecting the rooks. It also not only
applies pressure on the queenside but also keeps the e5–square in
check.
13.Rhe1
Going
for e5 is the most critical. Pushing h4–h5 is not so dangerous, the
easiest way to stop that is with Nh5. This is the strongest move
according to the engines.
13.Kb1
The
most often played move. If White starts with 13.h4 , it will likely
transpose after 13…Rac8 14.Kb1.
13...Rac8
It
is possible to offer the trade of bishops right away with 13…Nd7 ,
but Giri likes to bring the rook to the semi-open c-file first,
before deciding where to jump with the knight. The knight can
potentially also go to h5.
14.h4
There
is no real threat yet, but this move could be useful in the long run,
as now h4–h5 is always in the air.
14...Nh5
This
feels right, blocking the kingside offense, but in fact, the computer
has a point in proposing the counterintuitive 14…Nd7!? as well.
15.Bxg7
Kxg7
Black
has a comfortable position.
16.e5
This
is a clever attempt to bring Black off balance, but it doesn’t lead
to anything special for White.
16...Qxe5
17.Rhe1 Qf4
Offering
a trade of queens.
18.Qf2
Qb4 19.a3 Qc5 20.Qxc5 Rxc5 21.Rxe6 Re8 22.Ne4 Rcc8 23.Re1 Kf8 24.Ng5
h6 25.Nh3 Nf6 26.g4 Rc4 27.c3 Kf7 28.Kc2 ½–½ (28) Tirrell,D
(2228)-Zumbado,M (2159) ICCF 2025]
13...Nd7
Given
that White is not intending h4–h5 in the near future, it makes more
sense to keep the knight closer to the center and go for this move,
as opposed to Nh5.
14.Bxg7
Kxg7 15.e5
A
challenging push. Taking the e5–pawn would open the e-file and
Black would lose the crucial e6–pawn.
15...d5
Keeping
the center closed. Black is fine, especially if he keeps the Nb8–Nc6
maneuver in mind, while also keeping the e6–pawn under control with
Kf7 when necessary.
16.Kb1
16.a3
This
seems like the slightly more accurate way of preparing Ne2. In the
arising endgame, the king is possibly better placed on c1 than on b1,
as it is closer to the center. Although frankly, it is all very
subtle and the differences are minuscule.
16...Rac8
This
move never needs commentary in the Sicilian. The rook belongs on this
file!
17.Ne2
Qxd2+ 18.Rxd2
The
endgame is fine for Black. It is very important that, even if White
gets the grip with the pawn chain e5–f4–g3–h4, it is not the
end of the world, now that Black no longer has the bishop on g7.
White’s knight on d4 will be challenged with the thematic Nb8–Nc6!,
the king will likely reroute itself to d7 so as to stay close to
e6–pawn while letting the f8–rook have the view over the f-file.
And after that, black will keep the ideas of h6–g5 on the kingside
and the b5–a5 pawn advance on the queenside in mind.
18...a5
The
idea is to grab some space on the queenside.
19.Nd4
Kf7 20.Re3
If
21.f4 there is always the Nc5–Ne4 idea, which is the point of Black
keeping the d7–knight flexible.
20.h4
faced
the move h4 in one of my correspondence chess games.20...a4
21.Kb1 Rc4 22.b3 Rc3 23.Kb2 Rfc8 24.Ree2 h6 25.Rd1 Nb8 ½–½
(45) omni1 (2590)-turpaduunari (Christer Sundqvist) (2582)
https://gameknot.com/chess.pl 2025)
20...a4
21.Kb1
Again,
21.f4 is met with 21…Nc5!
21...g5
16...Rac8
17.Ne2
17.h4
This was played in a correspondence game. White goes for the h4–h5
push after all, now that Black's knight is far away from the
h5–square.
17...Nb6
More
forcing than 17…Rf5 . The Nc4–threat is strong now, so White has
to go for an endgame. If White ignores the threat, he will soon be in
trouble, for example after 18.h5 Nc4 19.Qc1 Nxb2 and Black is
crushing it.
17...Rf5
18.f4 Rc4 Black
increases the pressure and intends Nc5 next. White decided to go for
an endgame after all. 19.Ne2
Qxd2 20.Rxd2 Rf8 21.Nd4 Nc5 And
draw was agreed in a correspondence game ½ - ½ (21)
Molina,R-Mulder,C (1920) ICCF email 2017.)
18.Ne2
Qxd2 19.Rxd2 Nc4
Depending
on where the white rook will go, Black will either take on e5 (like
after 20.Rd3 Nxe5) or jump with the knight to e3 (after 20.Rdd1 Ne3
or 20.Rd4 Ne3).
17...Qxd2
18.Rxd2
18...Rg8
18...Kf7
Black
cannot take on e5 due to 19.Nd4 and White grabs the pawn on e6 with a
fork.
19.Nd4
Nb8
Black
is fine here, challenging White’s strong d4–knight and
potentially starting counterplay against the e5–pawn. A
correspondence game that reached this position (Efendiyev-Lecroq,
2017) was eventually drawn.
19.Nd4
Kf7 20.Re3 g5 21.Kc1 Nb8 22.Rb3 b6 23.Ra3 a5 24.f4 Nc6 25.Rh3 Nxd4
26.Rxd4 Rcf8 27.Rb3 Rb8 28.Rh3 Rbf8 29.Rb3 Rb8
½–½
(29) Broniek,M (2502)-Tauber,H (2528) ICCF 2024
