This thesis investigates the pricing of European options using Backward Stochastic Differential Equations (BSDEs) and deep learning. Explicit solutions for the pricing problem are only available in some particular cases like in the linear case under the Black-Scholes model, which does notreflect the reality of financial markets. Hence, to approximate the options’ prices, it is necessaryto use numerical methods. All traditional numerical methods for this problem are intractablein high dimensions. BSDEs offer a flexible alternative by reformulating pricing problems into aprobabilistic framework, linking stochastic calculus with partial differential equations via the Feyn-man–Kac formula. To overcome numerical challenges in solving BSDEs, this work implements the Deep BSDE method, which utilizes feed-forward neural networks to approximate the solution.The method’s performance is evaluated through extensive numerical experiments on both low-and high-dimensional options, including a 100-dimensional call spread with asymmetric interestrates. Results show that appropriate tuning of network depth, discretization, and training iter-ations significantly enhances accuracy. The study confirms the Deep BSDE solver’s effectivenessin complex financial settings, demonstrating its potential for robust option pricing under realistic market conditions