False recognition, or the mis-categorization of a "new" stimulus as "old," might support fixed false beliefs by blocking new learning or otherwise contributing to internal representations of the world that are at odds with reality. However, the mechanisms through which false recognition is facilitated among paranoid individuals remain unclear. We examined 2 phenomena that may contribute to this effect: an overreliance on fluency-based processes during recognition, manifesting as a lower threshold for judging items as recently studied, and a propensity to require less information to come to a highly confident judgment. The former would be expected to be particularly pronounced among items that are generally familiar, as opposed to completely novel. Here, we manipulated familiarity in a recognition memory paradigm by using stimuli that varied in their rate of extraexperimental exposure (i.e., real words vs. pseudowords). Further, to determine whether paranoia was associated with a tendency to differentially misallocate confidence to errors, we calculated a hierarchical Bayesian estimate of metacognitive sensitivity (meta-d') in addition to the more classic d'. In line with our hypotheses, paranoia was associated with an increased rate of false alarm errors, differentially so for familiar versus unfamiliar stimuli, suggesting that a context-agnostic, familiarity-based memory system might underlie observed memory distortions. What's more, paranoia was associated with heightened confidence on error trials and reduced metacognitive sensitivity. These findings highlight 2 distinct deficits-in both novelty detection and metacognitive monitoring-that contribute to false recognition judgments, offering targets for cognitive interventions to reduce memory distortion among paranoid individuals. (PsycInfo Database Record (c) 2021 APA, all rights reserved).