This empirical study investigates the effects of different caption modes on the content and vocabulary comprehension by Chinese English learners with different levels of English proficiency. The results show that the full captioned group performed better on content comprehension than the keyword group, while no significant difference was found on vocabulary comprehension between the two captioned groups. For the beginning-level learners, the full captioned groups did better both in content and vocabulary comprehension than the keyword caption group; meanwhile, for the advanced learners, both full captions and keyword captions similarly facilitated content and vocabulary comprehension. Therefore, the present findings suggest that keyword captioning is insufficient for foreign language learners' content comprehension, yet might be appropriate for their vocabulary learning. Furthermore, choosing the content caption mode for teaching EFL depends on students' English proficiency and their learning purpose.