This article presents two experiments using the DRM paradigm and examining the effects of variations in the depth of processing on the occurrence of false memories at recall. Contrary to what is generally observed, the results of the first experiment indicate that deep processing, maximizing the possibility of implication of distinctive characteristics, leads to an increase of the recall of hits without increasing the recall of lures. The second experiment uses instructions of inclusion requiring participants to recall not only the presented items but also all the words that were activated in memory both during the encoding and retrieval phases. These instructions, which deactivate the strategy of control of the source, support the notion that deep processing favours the activation of distinctive characteristics that facilitate the process of discrimination during the identification of the source. (PsycINFO Database Record (c) 2013 APA, all rights reserved). (journal abstract)