In this paper we present the prototype database system Concert and the incorporation of a framework of a generic index tree for spatio-temporal data. We show the ideas behind the Concert architecture as far as they are important to understand the framework approach presented. We show how the index is based on the conceptual behaviour of data in contrast to generalized algorithms or methods. Because of the simplicity of R-trees we take an R-tree like structure to explain our generic spatio-temporal index. It is remarkable that in Concert a generic index can be defined without any predefined “hard-wired” spatial or temporal data types such as intervals or rectangles. As it turns out the only important properties needed are an OVERLAP and a SPLIT function, the first one checking for spatial or temporal overlap of objects, the second one providing a hierarchical decomposition of the data space into subspaces. If, in addition, splitting of data objects is allowed we are able to define manageable node sizes, leading to an improved generic index similar to R+-trees or other derivations.