This paper examines the processes involved in the assignment of a pronominal anaphora to an antecedent in the scope of the universal quantifiers todo(s) and cada in Brazilian Portuguese. It investigates the preferential interpretations (collective or distributive) of the quantifiers and how this process affects the antecedent’s conceptual representation in terms of its number feature. The cost associated to the antecedent retrieval at the pronoun was different for each of the quantifiers. The results of the co-reference experiments are in accordance with previous studies and suggest that, while todos and cada respectively trigger preferential collective and distributive interpretations during real time processing, todo shows a less defined pattern.