Existing fake website detection systems are unable to effectively detect fake websites. In this study, we advocate the development of fake website detection systems that employ classification methods grounded in statistical learning theory (SLT). Experimental results reveal that a prototype system developed using SLT-based methods outperforms seven existing fake website detection systems on a test bed encompassing 900 real and fake websites.