Phishing is the act of using spoofed e-mails and fraudulent web sites to trick financial organizations and customers into revealing their personal or financial information. One of the main problems of phishing e-mail detection is the unknown “zero-day” phishing attack. A zero-day attack is one that phishers mount using hosts that do not appear in blacklists or using techniques that evade known approaches in phishing detection. Nowadays, phishers are creating different representation techniques to create unknown “zero-day” phishing e-mails to breach the defenses of detectors. This book proposes the Phishing Dynamic Evolving Neural Fuzzy Framework (PDENFF) that adapts the Evolving Connectionist System (ECoS) based online learning mode enhanced by offline learning mode. The proposed framework uses a hybrid supervised/unsupervised learning approach to speed up the system as well as to detect zero-day phishing e-mail attacks with a high level of accuracy and a low memory footprint. The proposed framework was tested, and a dynamic preprocessing and feature extraction system was implemented. MATLAB was used for the connectionist framework of the system engine as well as for computation.