Lexical bundles (LBs), as building blocks of fluent academic language production, have been studied extensively, but less attention has been paid to non-native authors' productions. Nor has there been a study to investigate the possible changes that LBs might undergo in non-native authors' productions when submitted to copy editors. Therefore, a 914,666-word corpus was selected, encompassing both edited and unedited articles in medical sciences written by Persian academics. Then, we studied the frequency of 4-word bundles alongside their structural and functional features, as well as the changes the editing process could bring about in this regard. The results indicated that although types and tokens of bundles generally increased by 6.6% and 17.3% respectively after the editing process, this increase did not include all subcategories. Furthermore, the editing process was not found to make any tangible structural or functional modifications. Pedagogically, our findings could have implications for non-native authors and editors.