The present study investigated whether or not native speakers of Japanese interpret null and overt pronouns as predicted by the Position of Antecedent Hypothesis after it is adapted to Japanese. The original hypothesis was proposed for Italian, and states that native speakers of a null subject language prefer to interpret null pronouns as referring to an antecedent in subject position and prefer to interpret overt pronouns as referring to antecedents in object position. These hypotheses have been studied in monolingual speakers of several other languages including Korean, Spanish, and Catalan, and it was revealed that the predictions of the PAH may not hold up across all languages. However, two limitations can be seen in the previous studies; languages investigated in most of the previous studies were Romance languages, and many of them employed only offline methods. Thus, the present study attempted to address both limitations by using an online method to see if the predictions of the PAH can be seen in the behavior of Japanese speakers when resolving anaphoric expressions. The present study adapted the PAH to Japanese, taking into consideration certain facts about case marking in Japanese. The adapted hypothesis predicted that the null pronoun and the overt pronoun with the topic case marker -wa would prefer a subject antecedent, whereas the overt pronoun with the nominative case marker -ga would prefer a non-subject antecedent. The online self-paced reading study with 36 native speakers of Japanese was conducted to address this issue. The results of the present study showed that all three pronouns (null pronoun, the overt pronoun with the topic case marker -wa, and the overt pronoun with the nominative case marker -ga) showed a preference for a subject antecedent in the critical region. This partially supports the PAH in that the null pronoun had a preference for a subject antecedent, but also challenges predictions of a difference between null and the overt pronouns. This suggests that the PAH, in its original or adapted form, does not make accurate predictions for Japanese.