In the context of cybersecurity, effective intrusion detection plays a crucial role in safeguarding computer networks and systems from malicious activities. The motivation for this project stems from the increasing complexity and sophistication of cyberattacks, which necessitates the development of advanced and accurate intrusion detection models. The aim of this work is to perform a comprehensive evaluation of Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN) for intrusion detection. CNN and RNN are two popular deep learning architectures known for their ability to extract meaningful patterns and temporal dependencies, respectively, making them suitable candidates for intrusion detection tasks. Two benchmark datasets: NSL-KDD and CICIDS2017 containing labeled network traffic data with various types of intrusions were employed and compared through multiple evaluation metrics. The results obtained from the experiments demonstrate the effectiveness of both CNN and RNN models in detecting intrusions. The CNN model achieved an accuracy of 86.40% on the NSL-KDD dataset and 95.20% on the CICIDS2017 dataset, while the RNN model achieved higher accuracy values of 96.20% and 94.10% on the respective datasets. Additionally, precision, recall, F1-score, error rate and other metrics were calculated and compared for both models. The results highlight the superiority of RNN in the NSL-KDD dataset and CNN in the CICIDS2017 datasets in terms of accuracy on the evaluated datasets. These findings contribute to the body of knowledge in the field of intrusion detection and can guide the selection and deployment of appropriate models for real-world applications, ultimately enhancing the security of computer networks and systems.