In this chapter, support vector machines (SVM) are introduced - a kind of classifiers developed specifically to achieve high predictive accuracy. First, the basic variant for binary classification into linearly separable classes is presented, which is then followed by extensions to non-linear classification, multiple classes and noise-tolerant classification. SVM are illustrated on examples from spam filtering, recommender systems and malware detection. In connection with SVM, the method of active learning is explained and illustrated on an example of SVM active learning in recommender systems.