The Deese/Roediger-McDermott (DRM) paradigm comprises the study of lists in which words (e.g., bed, pillow, etc.) are all associates of a single nonstudied critical item (e.g., sleep). The probability of falsely recalling or recognising nonstudied critical items is often similar to (or sometimes higher than) the probability of correctly recalling or recognising studied associates. False memories produced throughout this paradigm are usually seen as vivid, long lasting, and difficult to consciously avoid. Our experiment aimed to analyse the effect of dichotic listening and shadowing procedures on the production of false memories in the DRM paradigm. The results showed that the production of false memories under a divided attention condition during processing was not eliminated, independently of the type of memory task--recall or recognition. Moreover, the proportion of false memories produced in our study was similar to the amounts of correct recall and recognition in both encoding conditions. Therefore, our study confirms the robustness of this type of memory distortion (e.g., Gallo, Roediger, & McDermott, 2001), because even when the encoding conditions are impoverished, participants are prone to falsely remember the critical words, reinforcing the theoretical assumption of the automatic activation of critical lures.