The thermodynamic and structural properties of Tyrosine-Glycine-Leusine-Phenylalanine (YGLF, in a one letter code) and Lysine-Valine-Leusine-Proline-Valine-Proline-Clutamine (KVLPVPQ) peptide sequences were studied by three-dimensional molecular modeling in vacuum and solution. All the three-dimensional conformations of each peptide sequences were obtained by multicanonical simulations with using ECEPP/2 force field and each simulation started from completely random initial conformation. Solvation contributions are included by a term that is proportional to solvent-accessible surface areas of peptides. In the present study, we calculated the average values of total energy, specific heat, fourth-order cumulant and end-to-end distance for two peptide sequences of milk protein as a function of temperature. With using major advantage of this simulation technique, Ramachandran plots were prepared and analysed to predict the relative occurrence probabilities of beta-turn, gamma-turn and helical structures. Although structural predictions of these sequences indicate both the presence of high level of gamma-turns and low level of beta-turns in vacuum and solvent, it was observed that these probabilities in vacuum were higher than the ones in solvent model.