This study investigates the variability of native and nonnative raters in their assessments of Korean-speaking English (KSE) learners’ pronunciation with and without regards to the students’ English proficiency. Sixty 6th graders participated as speakers, and four KSE teachers and four native English-speaking (NES) teachers participated as assessors. The KSE and NES teacher ratings of the students’ English pronunciation in segmentals, stress-rhythm, and intonation were analyzed using t-tests and ANOVA. The results were as follows. The KSE teachers were stricter on the assessments of the students’ pronunciation in the three features across students of all English proficiency levels. The main effect of the KSE and NES raters on the assessments of the three features did not vary significantly according to the students’ English proficiency. The NES and KSE teacher ratings of the three features conformed to the students’ proficiency levels diagnosed by a written test. Both groups of raters tended to assess the accuracy of the suprasegmentals significantly higher than that of segmentals across students of all English proficiency levels, although the two assessor groups differed in the accuracy of segmentals and intonation for students of upper and intermediate levels. Based on the findings, implications for teaching and assessing English pronunciation were suggested.