ONTENTS

Contents

Symbols 4
Dedication 4
Acknowledgements 4
Bibliography 4
Introduction 5
Recommendations 10
Typical Ideas in the Marshall Attack 12

1 e4 e5 2 Ìf3 Ìc6 3 Íb5 a6 4 Ía4 Ìf6 5 0-0 Íe7 6 Îe1 b5 7 Íb3 0-0

Part 1: Main Lines with 8 c3 d5 9 exd5 Ìxd5 10 Ìxe5 Ìxe5

11 Îxe5 c6 12 d4 Íd6 13 Îe1 Ëh4 14 g3 Ëh3
1 Spassky Variation 20
2 Old Main Line: 18...f5/18...bxa4 31
3 15 Íe3: Other Lines 51
4 Modern Variation: 15 Îe4 74

Part 2: Other Lines after 8 c3 d5 9 exd5 Ìxd5

10 Ìxe5 Ìxe5 11 Îxe5 c6
5 Refined Rook-Lift: 12 d3 92
6 Elite Equalizer: 12 d3 Íd6 13 Îe1 Íf5 108
7 Early Deviations 120

Part 3: Anti-Marshall
8 Anti-Marshall: 8 a4 137
9 Anti-Marshall: 8 h3 155
10 Other Anti-Marshall Lines 170

Index of Variations 190

 OLD MAIN LINE: 18...f5/18...bxa4

2 Old Main Line: 18...f5/18...bxa4

1 e4 e5 2 Ìf3 Ìc6 3 Íb5 a6 4 Ía4 Ìf6 5 0-0

Íe7 6 Îe1 b5 7 Íb3 0-0 8 c3 d5 9 exd5 Ìxd5
-+-+-tk+
10 Ìxe5 Ìxe5 11 Îxe5 c6 12 d4 Íd6 13 Îe1 B +-+-+-zp
Ëh4 14 g3 Ëh3 15 Íe3 Íg4 16 Ëd3 Îae8 17
Ìd2 Îe6 18 a4 (D) p+pvr+-+
+-+n+p+q
-+-+-tk+ R+-Z-Zl+
B +-+-+pzp +LZ-V-Z-
p+pvr+-+ -Z-S-+-Z
+p+n+-+- +-+-TQM-
P+-Z-+l+
+LZQV-Zq methods to reach the Main Line should be
studied together because there are many recur-
-Z-S-Z-Z ring ideas that are good to know.
T-+-T-M- The variations in this chapter are very tacti-
cal in nature and the play is a real slugfest, with
both sides trying to get in the bigger and faster
Black is at an important crossroads. White punch. Theoretically, Black has been consid-
threatens 19 axb5 axb5 20 Íxd5 cxd5 21 Ëxb5. ered to be on rather shaky ground, but there is
Black can play 18...Ëh5 to protect d5, which still unexplored territory despite the line’s his-
was covered in Chapter 1. The alternative is to torical popularity in both over-the-board and
play for a direct attack on the kingside with correspondence play. In any case, this chapter
...f5. Black can begin this plan by playing ei- shows a lot of thematic ideas in the Marshall
ther 18...f5 or 18...bxa4, which often lead to Attack that will help Black find his way in other
the same position. The move-order that Black variations as well.
chooses should depend on the deviations for Section 2.1: The Old Road 18...f5?! 31
White that are possible after each move. The two Section 2.2: The New Road 18...bxa4 39
ways to reach the Main Line are 18...f5 19 Ëf1 Section 2.3: The Main Line 42
Ëh5 20 f4 bxa4 21 Îxa4 and 18...bxa4 19 Îxa4
f5 20 Ëf1 Ëh5 21 f4 (D). In both cases we reach
the position at the top of the next column. Section 2.1: The Old Road
It is from here that we form the basis for 18...f5?!
most of this chapter. The move-order that Black
chooses will allow White certain deviations, 1 e4 e5 2 Ìf3 Ìc6 3 Íb5 a6 4 Ía4 Ìf6 5 0-0
however. Some of these sidelines are danger- Íe7 6 Îe1 b5 7 Íb3 0-0 8 c3 d5 9 exd5 Ìxd5
ous, and some are not. Even though 18...f5 is 10 Ìxe5 Ìxe5 11 Îxe5 c6 12 d4 Íd6 13 Îe1
the traditional way of reaching the Main Line, Ëh4 14 g3 Ëh3 15 Íe3 Íg4 16 Ëd3 Îae8 17
I think that 18...bxa4 is the better way to head Ìd2 Îe6 18 a4 f5?! (D)
for the diagrammed position, so anyone want- Black ignores White’s demonstration on the
ing to play this variation would be well served queenside and threatens ...f4 and ...Îh6. If
learning this move-order. In any case, the two White meets a subsequent ...Îh6 with Ìf1,
32 UNDERSTANDING THE MARSHALL ATTACK

for Black. For example, 21...Íe2 22 Îxe2 Îxe2

-+-+-tk+ 23 bxc6 Îxd2 24 Íxd5+ Êh8 25 Ëf1 looks
W +-+-+-zp better for White. All is not lost, however, and 19
f4?! probably is a mistake in view of 19...bxa4!
p+pvr+-+ with the idea 20 Îxa4? Íxf4!, when 21 gxf4 is
+p+n+p+- met by 21...Îg6 and Black indeed has a strong
attack.
P+-Z-+l+ Thus we are left with two options for White.
+LZQV-Zq The first is the road White takes to head to the
Main Line, while the second is the very danger-
-Z-S-Z-Z ous ‘Internet Refutation’. We have:
T-+-T-M- A: 19 Ëf1
B: 19 axb5!?
32
37

then ...Íf3 comes and White no longer has Ëf1 A)

available to parry the mate threat on g2. Now,
the Main Line is 19 Ëf1 Ëh5 20 f4 bxa4 21 19 Ëf1 (D)
Îxa4, but there are a couple of other possibili-
ties, of which one is very important. Note that
19 Íxd5 cxd5 20 Ëf1 Ëh5 21 f4 will also
-+-+-tk+
transpose into main lines. B +-+-+-zp
After the poorly-timed 19 f4?!, the response
19...Îfe8 is always given as the refutation, but I p+pvr+-+
am not sure about this move. After 20 axb5
(worse is 20 Íf2? Íe2 21 Ëc2? Íxf4! as
+p+n+p+-
given by Nunn, one possibility being 22 axb5 P+-Z-+l+
Íxd2 23 Ëxd2 Íf3 24 Íe3 axb5! ø+) Black
plays 20...Íxf4 (D).
+LZ-V-Zq
-Z-S-Z-Z
-+-+r+k+ T-+-TQM-
W +-+-+-zp
This is the traditional move. White evicts the
p+p+r+-+ black queen and in doing so escapes potential
+P+n+p+- pins along the third rank and prepares to physi-
cally block the advance of Black’s f-pawn by
-+-Z-vl+ playing f4 himself.
+LZQV-Zq 19...Ëh5 20 f4
The only real alternative is 20 axb5 (D).
-Z-S-+-Z This was recommended by Shamkovich and
T-+-T-M- although it is not bad, it never really caught on.
Black has:
a) 20...axb5?! should favour White because
This has been considered to be virtually win- it will almost certainly help to have an open a-
ning for Black because, but after 21 gxf4? both file for the rook. 21 Íxd5 (21 f4 is also possible,
21...Îg6 and 21...Îh6 give Black a strong attack leading to the note to Black’s 20th move below,
while 21 bxc6? Íxg3! (better than 21...Îxe3? where Black avoids 20...bxa4, thus allowing 21
22 Íxd5+ Êh8 23 Îxe3 Íxe3+ 24 Êh1 Íxd2 axb5) 21...cxd5 22 Ëxb5 f4 23 Íxf4 Íxf4 24
25 c7!) 22 hxg3 Ëxg3+ leads to mate after ei- Îxe6 Íxe6 25 gxf4 Ëg6+ 26 Êh1 Ëc2 gives
ther 23 Êh1 Íf3+ or 23 Êf1 Ìf4!. However, Black some counterplay, but it is hard to believe
after 21 Íf2! I cannot find a good continuation that it is enough for the missing pawns.
 OLD MAIN LINE: 18...f5/18...bxa4 33

Black is pinned along the a2-g8 diagonal. It is

-+-+-tk+ an interesting situation where neither the d5-
B +-+-+-zp knight nor the e6-rook is pinned, but if one
piece moves, the other one will be. For this rea-
p+pvr+-+ son Black sometimes spends a tempo playing
+P+n+p+q ...Êh8 to free up these pieces.
20...bxa4
-+-Z-+l+ Black avoids the threatened 21 axb5 axb5 22
+LZ-V-Z- Íxd5 cxd5 23 Ëxb5, opens the b-file, and
lures White’s rook off the first rank. The a6-
-Z-S-Z-Z pawn is left to its fate, but Black hopes that he
T-+-TQM- will be able to generate enough kingside play.
To see why this move is supposed to be neces-
sary, let’s look at Black’s other sensible-look-
b) 20...f4 is a more sensible attempt to take ing moves:
advantage of White’s move-order and it is prob- a) 20...Îfe8?! (D).
ably stronger as well. 21 Íxf4 Íxf4 22 Îxe6
Íxe6 23 Ëe1!? (after 23 gxf4 axb5 Black has
compensation according to Nunn, and this as-
-+-+r+k+
sessment has held up in correspondence games) W +-+-+-zp
and now:
b1) 23...Íh3 24 Ëe4 (24 bxc6!? Êh8 25 p+pvr+-+
Íxd5 Ëxd5 26 Ëe4 may well be a better op-
tion) 24...Êh8 25 Ìf1 (25 bxc6 Ìxc3 26 bxc3
+p+n+p+q
Íxd2 27 Ëd3 Îxf2! 28 Êxf2 Íf5 leads to per- P+-Z-Zl+
petual check) 25...Îe8 26 Ëd3 axb5 27 gxf4
Ìxf4 28 Ìg3 Ëh4 gave Black compensation
+LZ-V-Z-
in McKenna-C.Chandler, corr. 1990-1. -Z-S-+-Z
b2) 23...Îe8 24 bxc6 Ìc7 looks rather un-
clear.
T-+-TQM-
We now return to 20 f4 (D):
Black tries to counter White’s attack on b5
and d5 by threatening the e3-bishop, but this
-+-+-tk+ natural move is probably just bad for tactical
B +-+-+-zp reasons. 21 axb5! axb5 (no better is 21...Îxe3
22 Îxe3 Îxe3 23 bxc6! Îe2 24 Íxd5+ and af-
p+pvr+-+ ter either 24...Êf8 or 24...Êh8 White will play
+p+n+p+q 25 h3! Íxh3 26 Íf3 leading to a winning end-
ing for White) 22 Íxd5 cxd5 23 Ëxb5 Ëf7
P+-Z-Zl+ (23...Íxf4 24 Ëxd5 is simply winning for
+LZ-V-Z- White, as pointed out by Nunn, while 23...Îxe3
24 Îxe3 Îxe3 25 Ëxd5+ also wins for White)
-Z-S-+-Z 24 Íf2! æ is given by Nunn. Black is just two
T-+-TQM- pawns down.
b) 20...g5?! (D) is a typical thrust that we
shall see again and again.
This move is a bit ugly positionally, but Black wants to break down White’s pawn-
something had to be done about the advance of chain on the dark squares at all costs. However,
Black’s f-pawn. On the plus side it gives White I think this move does not show enough respect
some more space and controls the e5- and g5- for White’s resources and, although compli-
squares. Although the e3-bishop is a bit loose, cated, I do not think this line holds up. 21 axb5
34 UNDERSTANDING THE MARSHALL ATTACK

and forced – Black must play for destruction.

-+-+-tk+ 24 Îxe3 f4 25 Îf3! Íxf3 26 Ëxf3 Ëxf3 27
W +-+-+-+p Ìxf3 fxg3 was Fridel-Elent, corr. 1996. White
can now play 28 Êg2 gxh2 29 Ìxh2 Íxh2 30
p+pvr+-+ Êxh2 Îf2+ 31 Êg3 Îxb2 32 Êf4 æ. The fire-
+p+n+pzq works are over and White has a big advantage
in the endgame. We shall see a similar version
P+-Z-Zl+ of this forcing play in other positions. Here it
+LZ-V-Z- just does not work, and this line looks like
enough reason for Black to avoid 20...g5.
-Z-S-+-Z c) 20...Êh8 (D).
T-+-TQM-
-+-+-t-m
axb5 22 Íxd5! (22 fxg5?! allows Black to es- W +-+-+-zp
cape after 22...f4! 23 Íxf4 Íxf4 24 gxf4 Îxf4
25 Ëd3 Íe2!, when White is forced to take p+pvr+-+
perpetual check by 26 Îa8+ Êg7 27 Îa7+
Êf8 28 Îa8+ Ó-Ó Roelens-Elent, corr. 1996)
+p+n+p+q
22...cxd5 (D) and now: P+-Z-Zl+
+LZ-V-Z-
-+-+-tk+ -Z-S-+-Z
W +-+-+-+p T-+-TQM-
-+-vr+-+
+p+p+pzq This line is also supposed to be dubious, but
maybe it is not so bad. By breaking the pins on
-+-Z-Zl+ the a2-g8 diagonal, Black threatens the e3-
+-Z-V-Z- bishop so White’s reply is forced. Black still
loses time and does nothing to address the
-Z-S-+-Z queenside problems, but White has not found
T-+-TQM- a convincing refutation yet. 21 Íxd5 (21 Íf2
Îh6 22 Ëg2 Íh3 23 Ëf3 Íg4 is a draw)
21...cxd5 22 axb5 and now:
b1) 23 Ëxb5?! Îh6 24 Ìf1 (after 24 Ëxd5+ c1) 22...Ëe8? 23 bxa6! (Black must always
Êh8 25 Ëg2? gxf4 26 Íxf4 Íxf4 27 gxf4 Íh3 be wary of sacrifices of this nature) 23...Îxe3 24
Æ Black’s attack is very strong) 24...Íf3 25 a7 Îxe1 25 Ëxe1 Ëxe1+ 26 Îxe1 Îa8 27 Îe6
fxg5 Íxg3 26 Îe2 f4 27 gxh6 Íxe2 28 Ëc6 Íc7 (St.Collins-P.Barrett, corr. 1999) and now
Íh4 29 Ëe6+ Êh8 30 Íxf4 Íf2+ 31 Êxf2 28 Êf2! (to stop ...Íe2) followed by 29 Îa6
Îxf4+ 32 Êg1 was drawn in Pietrocola-Elent, gives White a big advantage.
corr. 1999 because 32...Îxf1+! 33 Îxf1 Ëg5+ c2) 22...axb5 23 Ëxb5 (23 Ëg2!? is also
leads to stalemate after either 34 Êh1 Íf3+ 35 possible) 23...Îh6 24 h4! (24 Ìf1 Íf3 25 b3
Îxf3 Ëg2+ or 34 Êf2 Ëf4+ 35 Êxe2 Ëxf1+ g5! gave Black good play in Arias Duval-Gim-
36 Êxf1. enez, corr. 2003) 24...g5 25 fxg5 Íxg3 26 gxh6
b2) 23 Ëg2?! is murky after 23...Îfe8 24 Íh2+ 27 Êh1 and now instead of 27...Ëxh4?
Ëxd5 (24 Íf2 Ëf7 25 Îxe6 Îxe6 is also un- 28 Íg5! Ëxg5 29 Îe8 +ø Hage-Horak, corr.
clear) 24...Ëf7 25 Îa8 gxf4 26 Îxe8+ Îxe8 27 1999, Black should play 27...Íb8 28 Íg5 Íf3+
Ëxd6 fxe3. 29 Ìxf3 Ëxf3+ 30 Êg1 Ëg3+ 31 Êf1 Ëf3+
b3) By playing 23 fxg5!, White basically with a draw. This is a line that could be ex-
calls Black’s bluff. Now 23...Îxe3 is thematic plored further.
 OLD MAIN LINE: 18...f5/18...bxa4 35

21 Íxd5!?
21 Îxa4 is the Main Line, and will be con-
-+-+-tk+
sidered in Section 2.3. W +-+-+-zp
21...cxd5 (D)
p+-v-+-+
-+-+-tk+ +-+p+p+q
W +-+-+-zp p+-ZrZl+
p+-vr+-+ +-Z-V-Z-
+-+p+p+q -Z-S-+QZ
p+-Z-Zl+ T-+-T-M-
+-Z-V-Z- This interesting idea will probably not hold
-Z-S-+-Z up to modern (i.e., computer-assisted) scrutiny:
T-+-TQM- a) 23 Ìxe4?! fxe4 24 Îxa4 g5 25 Îxa6
gxf4 (not 25...Íf3? 26 Ëf2 Íxf4 27 Îa5! +ø)
and here:
22 Ëg2 (D) a1) 26 gxf4 and now 26...Îf6?! is line ‘b3’
White attacks the d5-pawn, which is rather below, but 26...Êh8! holds, since 27 Îxd6 (27
difficult to defend. Îea1 Îg8 28 Îa8 Íf8) 27...Îg8 28 f5 Íf3 29
22 Îxa4?? is a surprisingly common blun- Ëxg8+ Êxg8 30 Îd8+ leads to perpetual check.
der because of 22...Ëe8!. This is a typical tac- a2) 26 Îxd6 fxe3 27 Îxe3 Íh3 28 g4 wins
tical idea by which Black attacks the a4-rook a rook but not the game after 28...Ëh4 29 Ëxh3
and e3-bishop. White resigned immediately in Ëf2+ 30 Êh1 Îa8 31 Îd8+ Îxd8 32 g5 Îa8
Renet-Nunn, European Team Ch, Haifa 1989, 33 Ëe6+ and White must give perpetual check,
while 23 Íf2 Îxe1 24 Ëxe1 Ëxa4 25 Ëe6+ as has occurred in a few games.
Îf7 26 Ëxd6 Ëd1+ 27 Ìf1 h6 28 Ëd8+ Êh7 b) Therefore White tends to ignore the rook,
29 Ëxd5 Íh3 0-1 was Kindermann-Lukacs, at least for the time being: 23 Îxa4 g5 24 Îxa6
Budapest 1987. gxf4 and now:
b1) 25 Ìxe4?! fxe4 transposes to line ‘a’
above.
-+-+-tk+ b2) 25 Îxd6?! fxe3 (25...Îxe3? 26 Ëxd5+
B +-+-+-zp Êh8 27 Îxe3 fxe3 28 Ëe5+ is winning for
White, as pointed out by Nunn) 26 Ìxe4 fxe4
p+-vr+-+ transposes to line ‘a2’.
+-+p+p+q b3) 25 gxf4 is probably best. 25...Îf6 26
Ìxe4 fxe4 and now:
p+-Z-Zl+ b31) 27 Íf2 Îg6 28 Îxd6 (28 Íg3! looks
+-Z-V-Z- good for White) 28...Îxd6 29 Íg3 Íf3 30
Ëf1 Îg6 31 Êf2 is not so clear, as pointed out
-Z-S-+QZ by Nunn. Despite the two extra pawns, the op-
T-+-T-M- posite-coloured bishops and White’s draughty
king give Black reasonable chances.
b32) 27 h3! Ëxh3 (27...Îg6 28 Îxd6! Íe6
22...Îfe8 29 Îxe6 Îxg2+ 30 Êxg2 Ëf3+ 31 Êh2 leaves
This is the most sensible move, simply offer- White with way too much for the queen) 28
ing the d5-pawn, but Black has also tried to Ëxh3 Íxh3 29 Êf2 æ. If Black cannot improve
shield the pawn with the exchange sacrifice here (and he probably cannot), then 22...Îe4
22...Îe4 (D). has to be discarded.