This study investigated how English native speakers (ENSs) and Japanese native speakers (JNSs) at professorial, graduate, and undergraduate levels evaluate and edit ESL compositions written by Japanese college students. A total of 269 subjects first evaluated two compositions in terms of grammaticality, clarity of meaning, naturalness, and organization, using 10-point scales. English native speakers were more strict about grammaticality than were Japanese native speakers. In terms of clarity of meaning and organization, English native-speaking professors and graduate students gave more positive evaluations for both compositions than did the comparable Japanese-speaking groups. However, the Japanese undergraduates evaluated both compositions much more positively than did the English undergraduates. Comparisons in terms of naturalness were not generalizable because they showed different results between the two compositions. The subjects then edited the composition, correcting everything that seemed ungrammatical, unacceptable, or unnatural. ENSs provided far more corrections and corrected errors more accurately than did the JNSs. In both L1 groups, the higher the academic status of the evaluating group, the more accurately the group corrected errors. JNSs left many errors uncorrected, especially errors involving articles, number, prepositions, and lexical items which occur in Japanese as loan words from English.