Promoted by advanced experimental techniques for obtaining high-quality data and the steadily accumulating knowledge about the complexity of life, modeling biological systems at multiple interrelated levels of organization attracts more and more attention recently. Current approaches for modeling multilevel systems typically lack an accessible formal modeling language or have major limitations with respect to expressiveness. The aim of this thesis is to provide a comprehensive discussion on associated problems and needs and to propose a concrete solution addressing them. At first, several formal modeling approaches are examined regarding their suitability for describing biological models at multiple organizational levels. Thereby, diverse aspects are taken into account, such as the ability to describe dynamically changing hierarchical model structures and how upward and downward causation between different levels can be expressed. Based on the results of this study, a domain-specific language concept is developed to facilitate multilevel modeling in systems biology. The presented approach combines a rule-based modeling paradigm with dynamically nested model structures, attributed entities, and flexibly constrained reaction rates. Its expressive power, accessibility, and general usefulness for describing biological multilevel models are illustrated with the help of two exemplary case studies.