In this project, we focus on the ways how coreferential relations are realized in Czech in contrast to multiple languages. Coreference will be investigated in terms of expressions used for this purpose, their frequency and distribution. The nature of coreferential relations will be analyzed with respect to other linguistic aspects, e.g. tendency to use non-finite structures. Furthermore, we will perform several computational experiments employing the methods of coreference projection and bilingually-informed coreference resolution to measure how distant the languages are in terms of coreference and to support the findings of the theoretical analysis. The research will be carried out on multilingual parallel data that will be partially created within the project, including primarily Czech, English and Russian, with a possible extension to German and Polish. The key factor for such research on parallel data is the quality of word-alignment, which will be also addressed in this project.