This thesis aims to describe and explain the phase transitions of solvated alanine-poly-peptides using molecular dynamics simulations. It is shown that the influence of water is important to deliver biologically reasonable transition temperatures and to explain statistical distributions in the surface of backbone dihedral angles. Dynamic properties of the simulations elucidate α-helix formation and α-helix breaking. The method of statistical potential energy surfaces is presented, suggesting that the interaction of five consecutive residues determines the statistics of the peptides in the coiled state. A possible connection between statistical potential energy surfaces and protein folding is proposed.