Chessy
The Bishop's Opening: Thorold Gambit — Black's Revenge
White has sacrificed a bishop and a pawn to drag you into messy, tactical waters — but the engine says the gambiteer has overreached. Stockfish rates this position +2.13, a near-winning advantage for White. That means you are close to losing against perfect play. And yet, look at the human results: across 427 games Black actually wins 50.6% of the time, while White wins 46.6%. This is a gambit that works brilliantly in theory and crumbles under pressure in practice. Most White players do not find the precise continuation — and your job is to make them pay. Let's learn how.
Play the Bishop's Opening: Thorold Gambit against the engine
Free, no signup — you play black, the engine adapts to your level.
Ready to put the Thorold Gambit into action? Jump into the interactive drill below — play Black against an adapting engine and see if you can convert that 50.6%
Create a free account →What Black Is Playing For
The Thorold Gambit (1.e4 e5 2.Bc4 b5 3.Bxb5 f5) is a wild cousin of the King's Gambit. Black is not trying to hold the extra pawn — you already offered one back with 3...f5. Your ideas are straightforward: open the f-file for your rook, attack the centre with ...d5 in many lines, and develop quickly while White's light-squared bishop is off the board. The engine may say White is winning, but that evaluation assumes perfect follow-up. In a real game, White's king is still on e1, the f-file is about to open against them, and your pieces are ready to leap into action. You are the counter-puncher here.
The Engine's Top Move and How to Answer It
White's best response is exf5 (the only move that justifies the evaluation). After 4.exf5, the most-played follow-up is 4...Nf6 (developing with tempo, threatening ...Bc5 ideas and pressure on f5). The engine's full line runs: 4.exf5 Nf6 5.Nf3 c6. That 5...c6 move is important — it forces White's bishop to decide where to go and prepares ...d5 to reclaim the centre. From this position, practical play tends toward chaos, not sterile equality. Black scores well even here: White only wins 48.0% of the 298 games after exf5, meaning Black wins roughly half the time despite the engine's assessment. Trust the activity, not the number.
The Traps White Falls Into (Mistakes to Punish)
The statistics reveal that White players frequently choose safer-looking moves that are actually losing. Watch for these three gifts: d3 — this quiet developing move is an inaccuracy (losing about half a pawn compared to exf5). White scores just 46.7% from here. Bd3 — a full mistake that costs roughly 2.0 pawns. White scores only 38.9%. After Bd3 you can consider ...exf4 or simply ...Nf6 with an excellent position. Qh5+ — the most tempting bad move. It looks aggressive but loses about 1.3 pawns; White wins only 37.5% from here. Meet Qh5+ with ...g6, and after the queen retreats, your ...fxe4 gives you a dominant centre and a lead in development. If your opponent plays anything other than exf5, you are already better.
Why the Results Don't Match the Evaluation
This is a classic 'engine vs. human' opening. Stockfish sees that White can consolidate, develop, and cash in the extra piece later — but doing so requires precise moves under pressure. The Thorold Gambit scores 50.6% for Black over the board because White's position is harder to play than it looks. The f-file gets scary, the e4-pawn can become a target, and White's king safety is often neglected. Meanwhile, Black's plan is simple: develop, open lines, and attack. That imbalance in difficulty is why this gambit is a practical weapon — for Black. You get to steer the game toward tactical complications where your opponent is likelier to blink first. The engine's +2.13 is real, but so is Black's 50.6% win rate. Trust the latter.
Results across 427 Lichess games
| Most-played continuation | Games | White wins |
|---|---|---|
| exf5 | 298 | 48.0% |
| Nc3 | 30 | 43.3% |
| d3 | 30 | 46.7% |
| Bd3 | 18 | 38.9% |
| Qh5+ | 16 | 37.5% |
| Bc4 | 11 | 36.4% |
Frequently asked questions
Is the Thorold Gambit sound for White?
The engine says no — Stockfish rates the position after 3...f5 as +2.13, a near-winning advantage for White. That means, in theory, Black should lose against perfect play. However, in practice Black wins 50.6% of games, because White's winning plan is very hard to execute over the board.
What is Black's best move after 1.e4 e5 2.Bc4 b5 3.Bxb5 f5?
You do not play a move — it is White's turn. The most critical reply from White is 4.exf5, which is the engine's top choice. After that, Black's main response is 4...Nf6, developing a piece and preparing to challenge the centre.
What are the worst moves White can play in the Thorold Gambit?
According to the statistics, White's worst moves are Qh5+ (a mistake losing about 1.3 pawns; White scores only 37.5%), Bd3 (a mistake losing about 2.0 pawns; White scores 38.9%), and d3 (an inaccuracy losing about 0.5 pawns; White scores 46.7%). All are inferior to the best move exf5.
Should I play the Thorold Gambit as a club player?
If you enjoy chaotic, tactical battles where your opponent needs to find precise moves, go ahead. Just know that the engine says you should be losing with perfect defence. As Black, however, the Thorold Gambit is an excellent practical weapon — you are the one trying to punish White's most common mistakes.