Queen's Gambit Accepted: Rosenthal Variation with g3 — Playing as Black

In this line of the Queen's Gambit Accepted, Black has willingly given up the centre pawn on d5 in exchange for quick development and a solid pawn structure. Your main objectives here are straightforward: challenge White's centre with ...c5, develop your pieces to active squares, and castle kingside behind your pawn chain. The pawn White captured on c4 isn't a permanent gift — you'll often regain it or get excellent activity as compensation. The engine's small edge for White (+0.31) reflects that White has a lead in development and the bishop pair after playing Bg2, but your position is solid and free of weaknesses. Your plan is to put immediate pressure on d4, forcing White to make decisions about whether to hold the centre or let it go.

The computer's top recommendation is 5.Bg2, which appears in a massive 97,326 games. This natural developing move puts the bishop on the long diagonal, supporting the d4 pawn and preparing kingside castling. After 5...c5 6.0-0 Nc6, the game continues in a typical Queen's Gambit Accepted structure. The stats show White scores 57.4% from this position — solid, but nothing crushing for Black. You should be comfortable here: your ...c5 break has created tension in the centre, and you'll soon look to develop your light-squared bishop to b7 or d7, bring your rook to c8, and possibly play ...cxd4 to open the c-file for your rook. This is the main line, and knowing it well will carry you through many games.

One of the best things about playing this opening as Black is that White often goes wrong. Three of the alternative moves are clear inaccuracies that hand you an improved position. 5.Qa4+ (11,941 games) attacks your knight and pins the c4-pawn? Not quite — it drops about half a pawn in evaluation because after 5...Bd7 (or 5...Nc6) White's queen is misplaced and loses time. 5.Nc3 (5,287 games) is also an inaccuracy costing roughly half a pawn; it blocks the c2-pawn and gives you a comfortable game after 5...c5. 5.Qc2 (2,549 games) is even worse, losing about 0.8 pawns — the queen on c2 is vulnerable and doesn't help development. If your opponent plays any of these, you can seize the initiative with accurate replies. Only 5.e3 and 5.Ne5 are roughly okay, but they rarely appear and don't trouble Black.

The numbers from Lichess paint a clear picture. Across all 124,549 games reaching this position, White wins 56.4%, Black wins 39.5%, and only 4.2% end in draws. That low draw rate tells you something important: this opening leads to sharp, decisive games where both sides play for a win. As Black you're winning over a third of the time, which is respectable for a line where you're slightly worse out of the opening. Notice that when White plays the most popular move Bg2, their win rate is 57.4% — consistent with the overall average. But when White plays inaccuracies like Qc2, their score drops to 53.1%, and with e3 it plummets to 45.3% (where Black actually outscores White). This means you can turn the tables if your opponent doesn't know the theory.

When White plays 5.Bg2, the engine and database agree: your best reply is 5...c5, immediately challenging the d4 centre. After 6.0-0 Nc6, you've reached the tabiya of this variation. Your knight on c6 eyes the d4 square, your pawn on c5 creates tension, and you're ready to castle next. From here, typical ideas include completing development with ...Be7, ...0-0, and ...b6 (to develop the bishop to b7 or a6). You can also consider playing ...cxd4 on many moves, opening the c-file for your rook and forcing White to recapture with a piece. Just remember: you're the one who gave up the centre pawn early, so don't be shy about playing actively. Passive play is what allows White to consolidate their extra space.