French Defense: Mediterranean Defense (3.Nc3) – Playing as Black

The Mediterranean Defense (2...Nf6 before ...d5) is a flexible way to reach the French while sidestepping some of White's most dangerous systems. By delaying ...d5 one move, you avoid the Exchange Variation on move three and tempt White into committing their knight to c3 before you've committed your pawn to d5. In this position White has already played 3.Nc3, so your 3...d5 transposes back into a standard French — but with a key difference: your knight is on f6, not ...e7 as in the classical French. That knight is active early, pressuring d5 and preparing to recapture on d5 with a piece if White takes. The statistics back this up: across nearly 2.6 million games, Black actually wins 49.0% of the time, ahead of White's 47.4%. That tiny engine edge for White (0.44) hasn't translated into an advantage in practice.

Stockfish's top choice here is 4.e5, pushing the pawn and gaining space. White's idea is straightforward: cramp your pieces and then attack your knight with f4 and maybe g4 later. Your reply? The engine recommends the exact line 4...Nfd7, retreating the knight to its best square. From there you'll play ...c5 next, challenging White's centre. After 4.e5 Nfd7, White's most natural plan is f4, supporting the e5 pawn and preparing to develop with Nf3. This is the 'main line' for a reason — it's played over 1.5 million times in the database. As Black, you're aiming for ...c5 to blow open the centre and launch a queenside counterattack. Your knight on d7 is awkward for a moment, but it's well placed to support the ...c5 break.

While 4.e5 is the engine's top move — played in around 1.5 million of the 2.6 million database games — you will still face other moves in a large number of games, and some of them are downright inaccurate. Here are the three biggest gifts White can give you: 4.Bd3 is an inaccuracy worth about 0.7 pawns. White develops a piece but fails to gain space or challenge your knight — you can simply capture on e4 and emerge comfortable. 4.f3 is a full mistake, costing White about 1.1 pawns. This ugly move weakens the e1–h4 diagonal and delays development; strike immediately by taking on e4 and White loses control of the centre. 4.Nf3 is the worst of the bunch — a mistake costing roughly 1.5 pawns. White ignores the central tension and by capturing on e4 you end up with an extra pawn in a simplified position.

The raw numbers from the Lichess database are encouraging for Black. Let's look at how White scores against Black's Mediterranean Defense after 3...d5: 4.e5 (1.5M games): White scores 47.1% — that's below average for White, meaning Black is doing well. 4.Bg5 (529K games): White scores 49.8%, almost exactly even. 4.exd5 (184K games): White scores only 46.7%, below their usual performance. 4.Bd3 (156K games): White scores 46.9%, again below average. 4.f3 (70K games): White scores 45.6%, a miserable result for White. 4.Nf3 (34K games): White scores just 35.1% — almost a losing score for White at club level. The pattern is clear: the Mediterranean Defense is excellent for creating practical problems. Even the engine's preferred 4.e5 only holds White near parity. White's non-e5 moves all score poorly, and several of them are outright mistakes you can punish.