The binding of antigenic peptides to MHC class I molecules is an important step during the immunologic response of a host against a pathogen. Thus the prediction of binding sequences and the understanding of MHC/peptide(peptidomimetic) complex formation is crucial for vaccine design. We developed a new method, which combines sequence-based prediction with a new protein/peptide docking approach for the fast construction of bound conformations of predicted binders. The combined sequence and structure based algorithm is highly compute-time efficient and allows to screen large data bases of potential binders. Tested on the HLA-A0201 allele, the prediction method outperformed two other sequence-based methods used for evaluation. The docking algorithm was evaluated on 21 experimental MHC/peptide structures. The constructed peptide conformations could be refined within seconds to structures with an average backbone RMSD (peptide) of 1.53 Å (buried side chains: 1.12 Å, solvent exposed side chains: 2.3 Å) from the corresponding X-ray structures.