Abstract 　　　This study explored the the mechanism and collaborative facilitation in collaborative memory including two experiments. The first experiment investigated the influence of the size of the group and the patterns of retrieve to collaborative facilitation. The second experiment investigated the contribution of re-exposure and cross-cuing to collaborative facilitation. 　　　For the first experiment, a total of 336 undergraduates were randomly selected to participate in the present study. They were assigned to seven different conditions (individual group, pairs turn-taking collaborative recall, trios turn-taking collaborative recall, quartets turn-taking collaborative recall, pairs free-for-all collaborative recall, trios free-for-all collaborative recall, quartets free-for-all collaborative recall). Of these, 48 individuals were in each condition. Participants in this experiment studied a list comprising 8 categories. Instances were of low taxonomic frequency. Among of them, vegetables, fruit, trees and birds contained 6 instances, while animals, musical instruments, flowers and boats contained 7 instances. The experiment consisted of a 2 (retrieve patterns: free-for-all collaborative recall or turn-taking collaborative recall) × 3(group size: pairs, trios or quarters) between-participants design. Participants in individual group recalled individually. All participants studied the items individually. The items were presented one at a time for 4000ms each, with an interstimulus interval (ISI) of 250ms. Following the study, participants were asked to complete a 2-minutes mathematical calculate task to prevent rehearsal in short-term memory. Then participants were asked to recall the word either individually or in collaboration. After that, participants were asked to complete a 2-minutes mathematical game. Finally, all participants recalled the words individually. The results showed that no matter the group size is pairs, trios or quarters, retrieve pattern is free-for-all collaborative recall or turn-taking collaborative recall, the collaborative facilitation is present. Participates in the turn-taking group recalled more than those in the free-for-all group. The recall increased with the group size, and the participates in quarters recalled significantly greater than those in pairs and trios. 　　　In the second experiment , a total of 60 undergraduates were randomly selected. They were randomly divided in individual or collaborative group (pairs). There were 8 categories of low-frequency words, and each contained 7 instances. By installing an assistant, this study directly discussed the contribution of re-exposure and cross-cuing to collaborative facilitation. The assistant and a true subject recalled as a group. The assistant totally recalled eight words, each category one word. These eight words were studied but not recalled by the participants. The true subject recalled first for four minutes (i.e., reported all of his/her remembered), then the assistant continued to recall. If the subject came up with new word(s) while the assistant recalling, he or she also need to report. The experimenter recorded these response. The results suggested that re-exposure did contribute to collaborative facilitation, and greater than the role of cross-cuing. 　　　We have the following conclusions. No matter the group size is pairs, trios or quarters, retrieve pattern is free-for-all recall or turn-taking recall, the collaborative facilitation is present. The amount of collaborative facilitation of turn-taking recall is higher than free-for-all recall. The contribution of re-exposure to collaborative facilitation is greater than cross-cuing.