THE PEPTIDES THAT ARE EXPRESSED IN ASSOCIATION WITH MHC MOLECULES ARE VERY IMPORTANT IN THE PROCESS OF ANTIGEN RECOGNITION BY T CELLS. THE DETERMINATION OF AMINO-ACID SEQUENCES OF THESE T CELL EPITOPES IS OF GREAT VALUE IN THE VACCINE DESIGN. THE EXPERIMENTAL METHODS RELATED TO THIS TASK ARE VERY EXPENSIVE AND TIME CONSUMING. THEREFORE, SOFT COMPUTING TECHNIQUES ARE GETTING POPULAR TO REDUCE THE NUMBER OF PEPTIDES THAT SHOULD BE TESTED. IN THIS STUDY, ARTIFICIAL NEURAL NETWORKS ARE USED IN ORDER TO PREDICT THE EPITOPES. IT HAS BEEN SHOWN THAT HIGH SENSITIVITIES AND SPECIFICITIES CAN BE OBTAINED IN THE PREDICTION OF HLA-A*0201 ASSOCIATED PEPTIDES BY USING MULTI-LAYER PERCEPTRON NETWORKS. THE SENSITIVITIES AND SPECIFICITIES OF THESE NETWORKS ARE FOUND TO BE 91-92% AND 79-82%, RESPECTIVELY. CONSIDERING THE FACT THAT THE CORRESPONDING NUMBERS OF THE PREVIOUS REPORTS HAVE RARELY BEEN ABOVE 80%, THESE RESULTS ARE PROMISING.