Post-translational prote in modifications play an important role in many protein path ways and interactions. It has been hypothesized that modifications to prote insoccur in regions that are easily accessible, and many have been determined to belocated with in intrinsically disordered regions. However, identifying precise locations of prote in modifications involve sex pensive and time consuming laboratory work. Thus, automated identification of these sites is helpful. This paper studies methylated proteins and describes methods of building a predictor for arginine and lysine methylation sites using support vector machines. Our results indicate that, based on current data, both arginine and lysine methylation sites are likely to be intrinsically disordered and that the accuracies of methylation site predictions are high enough to be useful for prote in screening and for testing biological hypotheses.