This study reports on findings from two experiments investigating the interpretive patterns of overt pronouns in an embedded subject position with three types of matrix subjects (i.e. a referential NP, a quantified NP, or a "wh"-phrase) in Mandarin, English, and Japanese. According to the Overt Pronoun Constraint (OPC), overt pronouns in null-subject languages cannot have bound-variable interpretation, i.e. they cannot be bound by a quantified NP or a "wh"-phrase. This constraint has been assumed to be universal and accessible for learners at early stages of second language (L2) acquisition. The results of Experiment 1 show that, although Mandarin is a null-subject language, Mandarin and English native speakers as well as L2 learners of both languages demonstrated similar patterns of interpretation, accepting both coreference readings and bound-variable readings, the latter being contrary to the prediction of the OPC. The results of Experiment 2 show that Japanese native speakers differed from Mandarin native speakers in that the former accepted both coreference readings and bound-variable readings at chance levels. These findings clearly demonstrate that the OPC cannot be characterized as a property of null-subject languages generally, given the lack of effect in Mandarin, and there are cross-linguistic variations in interpretive patterns for overt pronouns among languages that exhibit the effect.