Many, if not all, languages possess syntactic constructions in which covert arguments, often referred to as “implicit arguments”, are required. While the implicit arguments in Western languages have received considerable attention, the study of the same topic in non-Western languages such as Japanese has been neglected. In this article, the nature of the implicit arguments of the potential and resultative constructions in Japanese is investigated. A detailed examination reveals that the implicit AGENTs in these two constructions differ in terms of the specificity of their reference, which in turn suggests that the representations of the two types of implicit arguments reflect this difference. It is argued that the implicit AGENT of the potential construction should be analyzed as proarb while that of the resultative construction is realized as pro.