The INSPIRE Directive requires data providers (e.g. Environmental Agencies or Mapping Authorities) from all European Member States to provide data conforming to a Europe-wide uniform data model, the INSPIRE Data Specifications. However, data from different sources can only be fully integrated in a certain destination schema, when the individual source schemas are identical. Otherwise the data need to be transformed semantically by means of schema translation so that they conform to the destination schema. Thus schema translation is regarded as the key to obtain harmonised data conforming to the INSPIRE Data Specifications. In order to be able to perform schema translation, the INSPIRE Directive proposes the so-called transformation services. With these services, data requested by a user can be directly transformed to conform to the INSPIRE Data Specifications, while the original data remains unchanged. All these issues are addressed by several research and development projects conducted at the GIS Group of the Technische Universität München since 2006. The projects focus on the integration of spatial data from the Lake Constance region where Germany, Switzerland and Austria share borders. In each country the data are stored in different systems, use different spatial reference systems and transfer formats and above all are based on different schemas determining the structure and semantics of spatial data. However, these differences do not only exist across the borders, but they also appear within each of these countries when data from topography and cadastre are used together (see supplement file). The projects use different approaches to define the transformation between the source schemas and the INSPIRE Data Specifications. They define the transformation rules not only on the level of data transfer formats (e.g. GML application schemas), but also on the platform-independent conceptual level (e.g. UML schemas) using a conceptual transformation language. The projects deal with issues such as: Are the data sources (spatial base data) enough to fill the objects, attributes and relations of the INSPIRE Annex I themes or must additional source data be provided to be able to completely generate the INSPIRE Annex I themes according to the data specifications? Are there problems regarding the semantic aspects of interoperability arising from the differing source and destination schemas? Do the different UML profiles of the source and destination schemas cause difficulties in defining the transformation? Is machine interpretability of the UML schemas currently fully provided, i.e., can UML schemas be used as input to transformation tools? How is transformation embedded in a web-based environment and which solutions for providing the transformed data through transformation services can be implemented? The presentation will present various projects and provide answers to the questions mentioned above. The problems arising from cross-border harmonisation are not specific to the region chosen in the projects; they are rather important to any region in which two or more countries share borders, for example in the case of Croatia.