This study compares the antecedent preferences of null and overt pronominal subjects in Italian and Spanish. With a series of self-paced reading experiments and questionnaire studies, Carminati (2002) showed that, in intra-sentential anaphora contexts in Italian, null subjects are read faster and interpreted more easily when they retrieve the most prominent antecedent available (defined in terms of syntactic configuration as the argument in the SpecIP position), while overt personal pronouns are processed more easily when they refer to a less prominent antecedent. The present study reports the results of a clause-by-clause self-paced reading experiment translated and run in Italian and Spanish with monolingual native speakers of each language. The results of the Italian part replicate Carminati's (2002) results; when the Spanish data are compared to the Italian they show that while the null subject seems to obey the same biases in both languages, cross-linguistic differences seem to emerge at the level of the overt pronoun. More precisely, the Spanish overt pronoun does not seem to be so strongly biased towards a non-prominent antecedent when compared to its Italian counterpart. The results are compatible with Accessibility theoretical assumptions (Ariel 1990) about cross-linguistic variation in the anaphoric preferences of referring expressions and are discussed in terms of Cardinaletti and Starke's (1999) cross-linguistic typology of pronouns.