Non-believed autobiographical memories [e.g. Mazzoni, G., Scoboria, A., & Harvey, L. (2010). Nonbelieved memories. Psychological Science, 21, 1334–1340] are striking examples of divergences between recollective experiences and beliefs in their correspondence to real events. After reviewing a broader range of similar phenomena, I argue that recollection–belief divergences can arise from normal, “healthy” metacognitive monitoring and control processes that balance memory recollections and reality constraints. Such validating “reality checks” draw on general world knowledge and external/social information. Importantly, changes in (perceived) reality constraints can lead to changes in memory beliefs. More generally, both recollection and (external) reality are keys to the past. In many cases, more or less automatic (System 1-type) reliance on recollection is sufficient (or memory would be useless as a system), but sometimes more elaborate (System 2-type) reality checks are needed. I conclude with some ideas about memory-driven and reality-driven recollection–belief divergences.