Abstract: Phishing, a term coined in 1996, is a form of online identity theft. Phisher tries to lure her victim into clicking a phishing URL pointing to a spoof page via spam-email to harvest financial information. The phishing activity is on the rise and their techniques become easier and more sophisticated. Quite a number of solutions to mitigate phishing attacks have been proposed to date. Those methods fetch webpage content which result in undesired side effects. In this paper, a novel method is proposed to detect phishing URL based on SVM. The feature vector is constructed with 23 features to model the SVM which 4 features are the structure feature of the phishing URL, 9 features are lexical feature and 10 features are mostly target phished brand name of website. The experimental results show the detection solution achieves 99.0% accuracy on average that the phishing URLs achieve is downloaded in PhishTank.