We evaluated the i-peptides occurrence frequency in the protein sequences, belonging to two reference datasets containing structured and disordered protein domains. Moreover we estimated the most frequent i-peptides (with i= 2, 3, 4) into these sequences in order to select specific i-peptides for each structural classification. According to these specific i-peptides, a new binary classification method was developed for predicting if a given protein sequence can be classified as "disordered" or "structured". The best results were obtained using the tri-peptides, much more able to gain structural information from sequences compared to the di-peptides.