Previous research has mostly approached face recognition and target identification by focusing on face perception mechanisms, but memory mechanisms also appear to play a role. Here, we examined how the presence of a mask interferes with the memory mechanisms involved in face recognition, focusing on the dynamic interplay between encoding and recognition processes. We approach two known memory effects: (a) matching study and test conditions effects (i.e., by presenting masked and/or unmasked faces) and (b) testing expectation effects (i.e., knowing in advance that a mask could be put on or taken off). Across three experiments using a yes/no recognition paradigm, the presence of a mask was orthogonally manipulated at the study and the test phases. All data showed no evidence of matching effects. In Experiment 1, the presence of masks either at study or test impaired the correct identification of a target. But in Experiments 2 and 3, in which the presence of masks at study or test was manipulated within participants, only masks presented at test-only impaired face identification. In these conditions, test expectations led participants to use similar encoding strategies to process masked and unmasked faces. Across all studies, participants were more liberal (i.e., used a more lenient criterion) when identifying masked faces presented at the test. We discuss these results and propose that to better understand how people may identify a face wearing a mask, researchers should take into account that memory is an active process of discrimination, in which expectations regarding test conditions may induce an encoding strategy that enables overcoming perceptual deficits.