What is the distinguishing feature of a living system that singularizes it from every non-living chemical ensemble, regardless of the extent of the complexity? The differentiable characteristic of the living system is information. Information assures the controlled reproduction of all the constituents, thereby ensuring the conservation of viability. Information – unlike energy – is not subject to a conservation law. Hence the fundamental question behind the origin of life is: How can information originate?Information theory, which was pioneered by Claude Shannon, cannot answer this question: this theory is most successful in dealing with problems of coding and transmission. In principle, the answer was formulated 130 years ago by Charles Darwin: The information that is unique for life evolves by virtue of natural selection. Today we can be more specific: natural selection is a non-equilibrium process. It is an inherent consequence of mutagenous self-replication at several levels of organization: for instance it is evident in molecules such as nucleic acids, in molecular complexes such as viruses and in autonomous formes of life such as micro- or higher organisms. New physical concepts have been introduced in order to deal quantitatively with the dynamics of the molecular generation of genetic information. They provide a physical foundation for Darwinian behaviour, yet they introduce major modifications in its interpretation. The lecture deals with these physical concepts, such as «sequence space», «quasi-species» and «hypercycles» and will scrutinize their adequacy for rationalizing experimental results obtained with molecular model systems and with viruses under natural conditions. Elucidating the principles of molecular self-organization has made possible to construct automated machines that make it possible for genetic information to evolve under controlled conditions in an abridged time scale.