In recent years, popular fora have seen lively discussion of the “Mandela Effect”. So called in reference to the paradigm case of a widely-shared apparent memory of Nelson Mandela’s death in prison in the 1980s, the effect occurs, roughly speaking, when individuals who have never met develop highly similar memories of events that never took place. Popular explanations of this phenomenon—e.g., that the seemingly inaccurate memories in question are in fact accurate memories of events that took place in parallel universes—are, toput it mildly, fanciful, and the academic literature so far contains little discussion of the effect or of the mechanisms that might be responsible for its occurrence. The goal of this chapter is to make a case for the existence of the Mandela Effect as a novel collective memory error worthy of serious scholarly scrutiny and to sketch a general account of the mechanisms that give rise to it. We argue, in particular, that the effect is an instance of collective confabulation, maintaining that this error occurs when individual misremembering goes online: whereas, in typical offline environments, subjects who give voice to mismemories about publicly-accessible events of the sort at issue in the Mandela Effect will usually encounter contradictory testimony, subjects who do so in the unusual environments constituted by certain online discussion fora may instead encounter confirmatory testimony, resulting in the reinforcement and stabilization of their mismemories and leading to convergence on shared but inaccurate representations of the past.