This paper documents the development of an empirically based system implemented in Prolog that automatically resolves several kinds of anaphora in Spanish texts. These are pronominal references, surface-count anaphora, one-anaphora and elliptical zero-subject constructions (i.e.sentences that omit their pronominal subject). The resolution is based on representations resulting from either partial or full parsing. The system developed can also work on the output of a POS tagger or with different dictionaries, without changing the grammar. This grammar represents the syntactic information of each language by means of the Slot Unification Grammar formalism. The different kinds of information used for anaphora resolution in full and partial parsing are shown, as well as evaluation results. The system has been adapted to English texts, obtaining encouraging results that prove that it can be applied with only a very few refinements to other languages as well as Spanish and English. In addition, the differences between English and Spanish anaphora are noted.