Korean [… V-ko iss-] is ambiguous between progressive and resultative meanings. We assume that it is ambiguous in three ways: P(rogressive)-reading, R(esultative)-reading and p(seudo)-R-reading constructions. These constructions allow different sets of predicates in the [… V] position. Most previous analyses agreed that telic predicates are responsible for the (traditional) resultative reading. However, this telicity condition does not adequately predict the occurrence of the reading: i) p-R predicates are not necessarily telic, and ii) not all telic predicates can trigger the R construction. The aim of this paper is to provide a set of constraints operating on the R construction under the understanding that p-R predicates should be excluded in advance. First, only those telic predicates that can indicate a present state in their past tense form can trigger the R construction. In addition, we show the need to posit two more constraints: a syntactic/semantic constraint and a pragmatic constraint. We further show that the present approach based on these new observations can properly characterize the R construction while excluding the P and p-R constructions. In dealing with the present state reading, we face the recalcitrant phenomenon of a theta role change in the subject of the predicate concerned.Keywords: [V-ko iss-] in Korean, resultative construction, telicity, past time interpretation, present state reading * I appreciate valuable comments from the three anonymous reviewers of this journal. I am also grateful to Ki-Sun Hong, Ilkyu Kim, Eunsuk Lim and Jae-mog Song for their constructive comments on various occasions.