DNA 5-hydroxymethylcytosine (5hmC), N6-methyladenine (6mA) and N4-methylcytosine (4mC) are three popular DNA modifications and involve in various of biological processes. Accurate genome-wide identification of 5hmC, 6mA and 4mC sites is invaluable for better understanding their biological functions. In this work, we initially proposed using K-tuple nucleotide frequency component, nucleotide chemical property and nucleotide frequency, and mono-nucleotide binary encoding scheme to formulate positive and negative samples. Subsequently, the Random Forest was utilized to perform the identification of 5hmC, 6mA and 4mC sites. Results of five-fold cross-validation test and independent dataset test showed that the proposed method could produce the good generalization ability, suggesting that our proposed method have good predictive performance to predict 5hmC, 6mA and 4mC sites. We anticipate that the iDNA-MS will become a powerful tool for large scale identification of DNA modification sites.