This paper focuses on perception of Japanese voiceless plosives by Korean listeners to: 1) assess whether the categorization of utterance initial- and medial- Japanese plosives into Korean three-way laryngeal categories is the same as in loanwords from Japanese; 2) identify the acoustic cues relevant to the identified patterns. We used modified stimuli from nonsense word tokens of native Japanese speakers, manipulating F0, amplitude and temporal characteristics. Loanwords represented only a part of the perception patterns: utterance initial [p] was mostly heard as aspirated and medial [p] was identified as either aspirated or fortis. The acoustic cues of F0 on the following vowel and temporal characteristics particular to the positions had effects on the laryngeal percept.