As the development of corpus linguistic analysis advances, lexical bundles, recurrent word sequences, have been recognized as crucial indicators of academic prose and conversation register. The current study investigates lexical bundle structures distributed in Korean and Japanese university students’ argumentative essays. 200,000-word corpora contributed by Korean and Japanese essays were built and four-word lexical bundles were automatically extracted. A series of statistical analyses of the lexical bundle frequencies confirmed that there were no significant differences in the distribution of the lexical bundle types. However, the Korean students use the lexical bundle tokens widely in academic prose. Conversely, the Japanese students use the lexical bundles in conversation. Implications of the findings and future research suggestions are discussed.