The aim of the study is to investigate whether hypermedia glosses (i.e., text only, text plus audio, and text plus video clips) promote EFL learners' reading comprehension and examine learners' preferences for using hypermedia glossing modes when they are reading a hypermedia text. The participants were 77 intermediate-level EFL learners from a university in China. Data were collected through pre- and post- reading comprehension tests, a questionnaire, and interviews. The reading comprehension findings reveal that all three glossing modes significantly promote learners' comprehension of the text, but no significant differences were found among these three types. In addition, the questionnaire and interviews data indicate that learners hold positive attitudes towards glosses and hypermedia reading. They preferred text coupled with video glossing mode more than text only and text plus audio glosses. The pedagogical implications were provided.