With the present research, the authors examined whether reminders could maintain 18-month-olds' memories generated from picture books and videos. Infants (N = 98) were shown a series of target actions in a picture book or on video. Either 24 hr or 2 weeks prior to a 4-week deferred imitation test, they were exposed to a reminder, a partial presentation of the original media demonstration. After both reminder delays, groups that received a video demonstration and a video reminder ("video/video") performed significantly better than did the video-reminder-only control group ("x/video"), but groups that received a picture-book demonstration and a picture-book reminder (book/book) did not perform better than did the picture-book-reminder-only control ("x/book"). Additionally, if reminders did not veridically match the conditions of encoding (e.g., video demonstration and a book reminder, "video/book" or vice versa), infants also failed to perform better than controls. Theoretical implications for the understanding of long-term memory processing during early childhood and practical implications for early multimedia usage are discussed.