The increasing complexity and uncertainty of modern manufacturing systems have significantly challenged traditional production planning and control methods. In particular, the Dynamic Job Shop Scheduling Problem (DJSSP) has emerged as a critical issue due to the presence of unpredictable events such as stochastic job arrivals, variable processing times, machine breakdowns, and frequent order changes. In this context, Reinforcement Learning (RL) has gained increasing attention as a promising paradigm for dynamic scheduling, thanks to its ability to model scheduling as a sequential decision-making process and to learn adaptive policies through interaction with the environment. This thesis investigates the application of reinforcement learning techniques to the Dynamic Job Shop Scheduling Problem, with the objective of assessing their effectiveness, robustness, and managerial relevance compared to traditional scheduling approaches.