The research demonstrated that there are differences in the processing of sentences containing anaphoric pronouns (e.g., "After Mark arrived, he used the phone.") and cataphoric pronouns (e.g., "After he arrived, Mark used the phone."). Reading time was measured on sentences containing 2 clauses; a pronoun appeared in 1 clause, and a proper name appeared in the other. The pronoun and name were either congruent or incongruent in gender. The results showed that anaphoric pronouns were resolved more rapidly than cataphoric pronouns when a co-referent interpretation was possible, but were resolved more slowly than cataphoric pronouns when a co-referent interpretation was not intended (i.e., in gender-incongruent conditions). The results of a follow-up experiment supported the conclusion that the observed processing differences stemmed from differences in readers' expectations for different pairs of referents (i.e., name-pronoun vs. pronoun-name) to be co-referent. "Expectation for co-reference" appears related to how quickly comprehenders use gender information either to accept or reject an interpretation of co-reference. Implications for theories of referential processing are discussed. (Contains 5 tables, 3 figures and 2 footnotes.)