Intuition suggests that “two heads together are better than two heads apart” when completing cognitive tasks, particularly problem solving. However, this does not appear to be the case for memory tasks. Researchers demonstrated that, during retrieval, individuals working together as a collaborative group performed much more poorly than did the same number of people recalling individually (nominal group). This phenomenon is called collaborative inhibition effect. There have been a number of studies on this topic since it was found two decades ago. Researchers have also given models to explain this effect. Psychologists suggested that collaborative inhibition might be due to the single theory such as retrieval inhibition mechanism, which suggests that when an individual engages in recall in a collaborative setting, relating the results of that recall to other group members might cause both the speaker and the listener to forget non-recalled materials, or retrieval disruption mechanism, which states that each individual’s idiosyncratic organization of overlapping information is disrupted during collaborative recall phases. However, there are growing evidences that it is inconsistent with the single theory account, suggesting that multiple mechanisms are needed to be explored underlying collaborative inhibition. To identify the contributions of retrieval inhibition and retrieval disruption underlying collaborative inhibition, the current study explored how the encoding mode and the encoding order consistency influence collaborative inhibition. Experiment 1 adopted a survival-processing paradigm to investigate whether retrieval inhibition plays a role in collaborative inhibition. In this experiment, 108 participants that compose 36 groups were randomly assigned to the survival-processing condition or non-survival-processing (control) condition. After the encoding phase, each group finished a group (collaborative or nominal) recall test. The results obtained here suggested that survival-processing condition and non-survival-processing condition both showed collaborative inhibition effect; but this deficit was attenuated in survival-processing condition, which means the encoding modes influenced the magnitude of collaborative inhibition, that indicated the retrieval inhibition mechanism plays a role in collaborative inhibition. Experiment 2 used associative memory training method to investigate whether retrieval disruption could influence collaborative inhibition. Another 120 participants were volunteered in this experiment. Half of the groups encoding items in the same order and the other half was not, both of whom were taught to encoding the materials with associative memory method in two weeks. The results of Experiment 2 showed that after the specific memory training, participants that learned items in the same order showed no collaborative inhibition, while participants that learned items in the different order showed classic collaborative inhibition. These results suggested that the encoding order consistency influence the magnitude of collaborative inhibition, which confirm the role of retrieval disruption mechanism underlying collaborative inhibition. Many results demonstrated that single theory (either retrieval inhibition or retrieval disruption mechanism) might be the mechanism of collaborative inhibition. However, the current results demonstrated that both retrieval inhibition and retrieval disruption mechanisms underlying collaborative inhibition. These results suggested that, compared to single theory, multiple mechanisms are more suitable to explain collaborative inhibition. Based on the above results, we discussed the mechanisms underlying collaborative inhibition. In addition, we give some important points to the future studies. Above all, collaborative inhibition may have multiple bases—in addition to retrieval inhibition, retrieval disruption also plays a role.