Molecular machines are essential components of living organisms. They are highly efficient and robust, much more than their macroscopic analogs. This stimulated growing interest in construction of artificial molecular machines with a set of functions which may be controlled in a specific way. Such man-made molecular complexes are designed as the building blocks for future nanotechnological devices. During the last decades many new molecular machines have been synthesized and characterized by various experimental techniques. This significantly increased our knowledge about systems of such kind and their functioning. However, there are only a few real applications of molecular machines. This is because the fundamental principles of operation of such single-molecule systems are not well understood. Existing theoretical studies, although very helpful, are still very sparse. This is because the molecular machines are very complex systems, comprising up to thousands atoms. Thus the progress in our understanding of nanoscale materials is tightly related to development of efficient computational and theoretical methodologies. In this work we studied two large classes of molecular machines: surface-moving nanocars and molecular rotors/motors, working on the surfaces and in crystalline state. In particular we studied the role of the internal interactions of these machines as well as their interactions with the environment. This included the flexibility of the molecules, including the rotation of the nanocars' wheels, effects of surface and rotors symmetry, charge transfer effects as well as many other factors. We have found out relations which determine the properties of studied classes of molecular machines. The development of computational and theoretical methods was another essential part of this work. In particular we have developed a family of the surface-molecule interaction potentials, aimed to performing long time scale and molecular simulations of complex systems. We also developed a physics-based model of the charge transfer happening between metals and the nanocars. This opened new ways to control such molecular machines. We also developed a theoretical framework to predict response of molecular rotors on various types of driving. Finally, we developed new and improved existing rigid-body molecular dynamics methods and extensively used them in our studies of molecular machines.