Today, the banking sector has been a target for many phishing attackers. The use of email as an electronic means of communication during working hours and mostly for official purposes has made it a lucrative attack vector. With the rapid growth of technology, phishing techniques have advanced as seen in the millions of cash lost by banks through email phishing yearly. This continues to be the case despite investments in spam filtering tools, monitoring tools as well as creating user awareness, through training of banking staff on how they can easily identify a phishing email. To protect bank users and prevent the financial loses through phishing attacks, it important to understand how phishing works as well as the techniques used to achieve it. Moreover, there is a great need to implement an anti-phishing algorithm that collectively checks against phishing linguistic techniques, existence of malicious links and malicious attachments. This can lead to an increase in the performance and accuracy of the designed tool towards detecting and flagging phishing emails thus preventing them from being read by target. Evolutionary prototyping methodology was applied during this research. The advantages are in the fact that it enabled continuous analysis and supervised learning of the algorithm development until the desired outcome was achieved. This research aimed at understanding the characteristic of phishing emails, towards achieving defence in depth through creation of an algorithm for detecting and flagging phishing emails. In this research, we have implemented a client-based anti-phishing algorithm. The algorithm is able to analyse phishing links, identify malicious email attachments and perform text classification using a Naïve Bayes classifier to identify phishing terms in a new unread email. It then flags the email as malicious and sends it to the spam folder. Therefore the user only gets clean emails in the in-box folder