Phishing attacks are one of the most common social engineering attacks targeting users emails to fraudulently steal confidential and sensitive information. They can be used as a part of more massive attacks launched to gain a foothold in corporate or government networks. Over the last decade, a number of anti-phishing techniques have been proposed to detect and mitigate these attacks. However, they are still inefficient and inaccurate. Thus, there is a great need for efficient and accurate detection techniques to cope with these attacks. In this paper, we proposed a phishing attack detection technique based on machine learning. We collected and analyzed more than 4000 phishing emails targeting the email service of the University of North Dakota. We modeled these attacks by selecting 10 relevant features and building a large dataset. This dataset was used to train, validate, and test the machine learning algorithms. For performance evaluation, four metrics have been used, namely probability of detection, probability of miss-detection, probability of false alarm, and accuracy. The experimental results show that better detection can be achieved using an artificial neural network.