The regulation of gene expression in cells is a stochastic many body process. An important source of complexity in the interactions between genes lies in the molecular details, which control the properties of small genetic networks. The relatively small number of protein molecules of a given type present in the cell and the nonlinear nature of chemical reactions results in complex behaviours which are hard to predict from first principles. I will discuss mathematical models and approximations which allow for analytical progress in studying noise on different levels of the regulatory system. I will show examples of how molecular noise can influence the cell's phenotype and together with information flow considerations lead to predict not only the connectivity but also the detailed biochemistry of a biological network. Lastly, I will discuss different approaches of how a stochastic molecular level description can be successfully expanded to larger regulatory systems.