A variety of evidence demonstrates that memory is a reconstructive process prone to errors and distortions. However, the complex relationship between memory encoding, strength of memory reactivation, and the likelihood of reporting true or false memories has yet to be ascertained. We address this issue in a setting that mimics a real-life experience: We asked participants to take a virtual museum tour in which they freely explored artworks included in the exhibit, while we measured the participants’ spontaneous viewing time of (...) each explored artwork. In a following memory reactivation phase, participants were presented again with explored artworks, followed by novel artworks not belonging to the same exhibit. For each of these objects, participants provided a reliving rating that indexed the strength of memory reactivation. In the final memory recognition phase, participants underwent an old/new memory task, involving reactivated vs. baseline targets, and activated and baseline lures. The results showed that those targets that were spontaneously viewed for a longer amount of time were more frequently correctly recognized. This pattern was particularly true for reactivated targets associated with greater memory strength. Paradoxically, however, lures that were presented after targets associated with higher reliving ratings in the reactivation phase were more often erroneously recognized as artworks encountered during the tour. This latter finding indicates that memory intrusions, irrespective of the viewing time, are more likely to take place and be incorporated into true memories when the strength of target memory is higher. (shrink)