•Vide-dependent recognition is based on recollection or familiarity processes.•Video context mainly functions as a global context and sometimes as local context.•Production mechanisms are influenced by context load and method of presentation. The present study investigated how video-dependent recognition is influenced by context load (number of items presented per context), the sequence of context presentation (successive or shuffled presentation of short video contexts), and item cue strength (length of study time). In addition, the effects of background photographs selected from still images of the video contexts were also investigated. The present study includes 7 experiments, in which a total of 287 undergraduates intentionally studied a list of unrelated words. After a filled 5-min retention interval, participants received a recognition test. In the same-context condition, the same video as at study was presented at test, whereas in the different-context condition, new videos were presented at test. With a context load of 1 item, a concordant effect and context dependent-recognition discrimination were found in two video experiments and one background-photograph experiment. Successive presentation of 4-s video contexts produced context-dependent effects in hit rate and in d' with a context load of 6 items and study time of 4 seconds per item, and with a context load of 18 items and study time of 1.33 (4/3) seconds per item. In contrast, no effect was found in the false alarm rate in either condition. Nonetheless, when the context load was 18 and the study time was 4 s per item, no effect was found for the hit rate, false alarm rate, or d'. Shuffled presentation of 4-s video contexts produced a concordant effect but no effect for d' with a context load 6 or 18. Between-participants manipulation produced a context-based mirror effect. The present results imply that video context functions mainly as a global environmental context, and sometimes as a local environmental context.