miRPara, is a new software tool that predicts mature miRNA in a species specific manner based on three SVM trained models from a comprehensive set of different parameters related to the physical properties of the pre-miRNA and its miRNAs. It achieves an accuracy of up to 80% for animal and virus sequences against experimentally verified mature miRNAs predicted from long genome sequences, making it one of the most accurate methods available. Because of the greater diversity of the plant miRNAs, only 70% accuracy was achieved, but this was nevertheless more accurate than results obtained with other prediction software.