Sicilian Najdorf: a White repertoire with the English Attack and the quieter 6.Be2

Short answer

The Najdorf arises after 1.e4 c5 2.Nf3 d6 3.d4 cxd4 4.Nxd4 Nf6 5.Nc3 a6, and White's sixth move decides the kind of game ahead. This guide closes a whole White repertoire: the English Attack with 6.Be3, the cheaper 6.Be2, what to do against 6...Ng4, against 6...e6 and against the deviations that turn up at the club, plus the position where the most thematic move in the Sicilian goes from best to worst because of two pawn moves. Every evaluation was measured with Stockfish 19 at depth 22, from the lines published here.

The essentials in a few lines

• No White sixth move exceeds +0.35 in the Najdorf tabiya: 6.f3 measures +0.35, 6.Be3 +0.31, 6.h3 +0.26, 6.Bg5 +0.25 and 6.Be2 +0.24. The choice is about the kind of game you want, not about which move refutes the defence.
• After 6.Be3 e5 the d4-knight is attacked by a pawn, and the English Attack's routine move becomes a piece blunder: 7.f3 measures −3.71, against +0.39 for 7.Nb3.
• Against 6...Ng4, defending the e3-bishop with the queen is the reflex that gives the opening away: 7.Qf3 measures −0.10 and 7.Qd2 −0.09 — both hand back more than the whole edge — against +0.32 for 7.Bg5 and +0.30 for 7.Bc1.
• The Nd5 jump is worth whatever the concrete position allows, not what its reputation promises: after 12...Ne8 it measures +1.95 and decides the game; after 12...Nh5 it measures +0.25 and is only the third best move, behind 13.Kb1 (+0.33) and 13.a3 (+0.32); and after 12...b4 it measures −0.11 and is the worst move available, against +0.44 for 13.Ne2.
• 6.Be2 costs four hundredths of a pawn against the English Attack (+0.19 after 11.a4, versus +0.23 in the 12.g5 tabiya) and skips the part of the theory that changes most often — it is the repertoire for two hours a week.

The tabiya after 5...a6, and what the pawn buys

The Najdorf is said to demand twenty memorised moves, and the club player without them is supposedly lost before the game starts. The tabiya says otherwise. Measuring the seven most played White sixth moves at depth 22, the best of them, 6.f3, gives +0.35; 6.Be3 gives +0.31; 6.h3 gives +0.26; the venerable 6.Bg5, once the main line of the world, gives +0.25; and 6.Be2 gives +0.24. Twenty-three hundredths of a pawn separate best from worst. None of them refutes anything.

What the a6-pawn buys is concrete and small. It rules out Nb5 and Bb5 — in other Sicilians precisely the moves that harass the d6-pawn — and prepares ...b5 to take space on the side where Black holds a majority. The price is a move: while Black plays a6, White develops. That move is the capital White plays the whole opening with, and spending it badly — on an unnecessary pawn move, on an early queen sortie — is what turns +0.31 into equality with nothing dramatic having happened.

This guide closes two choices: 6.Be3, the English Attack, with queenside castling and a pawn advance, and 6.Be2, cheaper to maintain. The sections that follow also cover 6...Ng4 and 6...e6, and the deviations that turn up on a Saturday afternoon — which is where a real repertoire parts company with a memorised line.

What ...e5 builds, and what it hands over

After 6.Be3 e5 something happens that the diagram shows and autopilot does not: the d4-knight is attacked by a pawn. Anyone following the memorised English Attack sequence and playing 7.f3 there loses the piece on the spot — the engine measures −3.71, and the game is over on move seven. It is the cheapest blunder to avoid and the most expensive to commit in this entire guide, and it is not hypothetical: f3, Qd2, O-O-O is exactly the sequence one learns as a block.

The three legitimate retreats were measured. 7.Nb3 gives +0.39 and has no serious rival: the knight lands where it watches a5 and c5, the two squares Black's queen and pawn want. The other two plausible retreats cost nearly the whole opening edge — 7.Nf3 gives +0.10, handing the tempo back for Black to hit the knight again, and 7.Nde2 levels the game outright (0.00), guarding d5 with one more piece at the cost of blocking the f1-bishop. 7.Nf5, which looks aggressive because it plants the knight in front of Black's king, measures −0.50: the square is pretty, the knight has no support and ...d5 comes with tempo.

The d5 hole is the permanent legacy of ...e5, and the whole Najdorf turns on who gets to use it. For White it is worth something once a knight lands there and Black's light-squared bishop has been traded; for Black it is irrelevant as long as the e6-bishop can capture whatever arrives. That is why White plays to trade that bishop, and it accounts for half the moves in the next two sections.

The English Attack, move by move, to the decision point

The reference line is 6.Be3 e5 7.Nb3 Be6 8.f3 Be7 9.Qd2 O-O 10.O-O-O Nbd7 11.g4 b5 12.g5. Every White move does two jobs, which is what keeps the plan learnable without becoming rote: f3 holds e4 and frees g4; Qd2 connects the rooks and backs up Be3; castling long takes the king off the file that is about to open. The order of f3 and Qd2 is the only point that gets argued about, and it matters less than advertised — with 8.f3 Black's best reply measures +0.32, with 8.Qd2 it measures +0.30. Two hundredths of a pawn, which is less than the measurement's own noise.

Black's replies were measured at three points. On move eight, 8...Be7 (+0.32) is the most accurate, ahead of 8...Nc6 (+0.35) and 8...h5 (+0.38), and well ahead of 8...Qc7 (+0.51). On move ten, 10...Nbd7 (+0.37) is again the best, and 10...b5 (+0.70) already costs three tenths, while 10...h5 — the move that stops White's advance before it starts — measures +0.90, because it freezes Black's kingside without solving the centre. On move twelve, after 12.g5, only 12...Nh5 (+0.23) holds the position: 12...b4 (+0.51) already concedes half a pawn.

This is where the most famous move in the Sicilian shows it has no opinion of its own. After 12...Nh5, the 13.Nd5 jump measures +0.25 and is only the third best move in the position, behind 13.Kb1 (+0.33) and 13.a3 (+0.32) — two waiting moves nobody memorises — and ahead of just 13.Rg1 (+0.17). After 12...b4, with the same pieces and two pawn moves' difference, the very same 13.Nd5 measures −0.11 and is the worst move available: 13.Ne2 measures +0.44. The knight on d5 is neither good nor bad by nature — it is good while Black's pawn is still on b5 and the c3-knight has nowhere to go without losing time.

6...Ng4: the move that asks what the bishop is doing on e3

6...Ng4 shares the top spot against 6.Be3 with 6...e5 (+0.36 to +0.35, the same number for practical purposes) and is the more awkward of the two, because it hits the piece that has just moved and asks whether it had anywhere to stand. The reflex of anyone taught that an attacked piece must be defended is Qf3 or Qd2, and it is exactly that reflex the engine punishes: 7.Qf3 measures −0.10 and 7.Qd2 −0.09. Defending a piece with a more expensive piece, in a position where the opponent can trade it off at will, is paying dearly for a problem that a bishop move solved.

7.Bg5 is the main reply and the ambitious one. After 7...h6 8.Bh4 g5 9.Bg3 Bg7 the position measures +0.38: Black has taken kingside space, but spent three pawn moves doing it, and that king will have to live somewhere. The alternatives on Black's ninth are the two worst — 9...Nc6 (+0.74) and 9...Ne5 (+0.84) — which says the line is rather more dangerous for an unprepared Black than its reputation suggests.

7.Bc1 looks like an admission of error and is not. The bishop goes home, Black has gained a move, and the position still measures +0.30 — practically the same as 7.Bg5. Black's best is 7...Nf6 (+0.23), offering a repetition. For anyone unwilling to study the 7.Bg5 line it is a defensible choice, with one honest condition: you must be willing to take the draw by repetition against an opponent equally unwilling to commit.

6.Be2: less theory, the same game

The line is 6.Be2 e5 7.Nb3 Be7 8.O-O O-O 9.Be3 Be6 10.Qd2 Nbd7 11.a4, and the plan fits in one sentence: stop ...b5 with a4, trade the e6-bishop for a knight on d5, and play the endgame with the hole. After 11.a4 Black's replies measure between +0.19 (11...Nb6) and +0.33 (11...Qc7) — a small, stable edge, the kind converted by technique rather than by attack.

White's ninth move is where the line shows its nature. 9.Be3 measures +0.27, 9.Qd3 +0.18, 9.Re1 +0.14 and 9.Bf3 +0.10 — seventeen hundredths of a pawn from first to last, with the middle two tied inside the measurement's noise. There is no single right move to memorise there, and that is the practical advantage: players of 6.Be2 make fewer mistakes because they get fewer chances to make one. The matching drawback is that the opponent gets fewer too — the positions are clear to both sides.

There is one trap in this line, and it is the kind that decides club games. After 8.O-O the e4-pawn looks unguarded, and ...Nxe4 looks like winning a pawn with tempo. It measures +5.73 for White: the c3-knight was never pinned, and Black ends up a piece down. Knowing the number is what keeps you from hesitating when it appears with the clock running.

Club deviations: when Black leaves the script on move six

The commonest club deviation is 6...b5, played too early for the same reason ...a6 was played: it looks as though Black's plan is a queenside advance, so the sooner the better. But ...b5 before ...e5 leaves the d5 square and the a4-e8 diagonal without a guard. Measured, the two candidates give 7.Nd5 +0.98 and 7.a4 +0.77 — close to a full pawn, twice what the main lines in this guide deliver, and won without a single tactical move. The order between them matters: occupying the square the opponent has just abandoned is worth more than attacking the pawn that abandoned it.

The other three are milder and follow the same logic. Against 6...Nbd7, 7.Be2 measures +0.66; against 6...g6, a Dragon with ...a6 thrown in, 7.Be2 measures +0.74; against 6...Nc6 the position is already worth +0.45 before White does anything. None of them has a two-move refutation, and hunting for one is the shortest way to hand the edge back — the punishment is positional, and comes from the opponent having spent time.

The practical rule that falls out of this holds for the whole repertoire: when the opponent leaves theory early, the answer is almost never the most aggressive move on the board. It is the move you were going to play anyway, made in the right order. The table below collects the four replies with their numbers, for reading before a tournament.

6...e6: the Scheveningen through the back door

6...e6 is Black's third answer to 6.Be3 by evaluation (+0.48) and the solidest of the three: instead of shutting the centre and living with the d5 hole, Black keeps a pawn on e6 guarding the square, at the price of locking in the c8-bishop. For White that is good news disguised as a complication, because the attacking plan stays the same, now without a closed black centre to compensate.

The main line is 7.f3 b5 8.Qd2 Nbd7 9.O-O-O, and there Black needs the right prophylactic move: 9...h5 measures +0.26, while 9...h6 (+0.46) and 9...Bb7 (+0.52) let White's attack roll. Here, unlike everywhere else in the guide, the repertoire's choice does cost something: 7.Be2 measures +0.52 and 7.f3 +0.35. Keeping one plan for both structures is still worth it for the study time saved, but the price is seventeen hundredths, and it is worth knowing it exists.

The mistake that actually shows up is the early short castle. After 7...Be7 8.Qd2 O-O 9.O-O-O, the measured black continuations sit between +0.74 and +1.08 — the black king has moved to precisely the wing White's pawns are heading for. It is the same misunderstanding as the ...h5 in the third section: the move looks prudent because it is the prudent move in almost every other opening.

Three weeks of study, and what to measure afterwards

Order matters more than volume here. A club player meets 6...b5, 6...Nbd7 and 6...g6 far more often than the 12.g5 tabiya, and it is precisely the deviations nobody studies — on the grounds that they are easy. Starting there means the repertoire pays off in the first week, and that payoff is what sustains the second week's study.

The part that truly needs memory is short: the three moves of 7.Bg5 h6 8.Bh4 g5 9.Bg3 against 6...Ng4, and playing f3 before g4 in the main line. Everything else is plan rather than sequence — and a plan is learned by playing four long games with it and reviewing each one, which fits into a week without taking a whole evening.

After three weeks, what tells you the repertoire has taken hold is not the results: it is where the errors appear. If your error points against the Najdorf sit on move 7 — the d4-knight, the queen coming out early — the problem is move order, and rereading the second section fixes it. If they sit on move 20, with the position already set up, the problem is middlegame planning, and the repertoire is not to blame. That distinction is worth measuring before switching openings, because switching openings never fixes the second kind.

• Week 1 — the deviations: 6...b5 (7.a4), 6...Nbd7 and 6...g6 (7.Be2), with four long games playing nothing else
• Week 2 — the main line: 6.Be3 e5 7.Nb3 Be6 8.f3 Be7 9.Qd2 O-O 10.O-O-O, until you can explain each move's job without looking
• Week 3 — 6...Ng4 and 6...e6: the two replies that change the type of position, and the only ones needing a memorised sequence
• If you have two hours a week or less, swap the English Attack for 6.Be2 and go straight to week 3

Frequently asked questions

Openings mentioned in this guide

• Sicilian Najdorf (B90)
• Sicilian Defense (B20)
• Sicilian Dragon (B70)