Two experiments investigate how people create spatial mental models (SMMs) as a function of presentation format. Spatial representations following a route within a virtual environment (visual) and an auditory description (verbal) were compared. Moreover, a dual-task procedure was adopted to depict the role of working memory (WM) components in such a task. Two testing formats were examined in both experiments: sentence verification (verbal) and sketch map (visual) tasks. This study shows that both presentations lead to adequate spatial representation. Nonetheless, SMMs are not presentation free. They are built by virtual presentation with a contribution of visuo-spatial WM. An involvement of verbal WM emerged while encoding spatial information from both verbal and virtual presentation. Congruency between presentation and testing formats slightly reduces the gap between verbal and virtual presentation. SMMs appear to be affected by the interaction of presentation and testing format as well as by the WM availability. Copyright © 2013 John Wiley & Sons, Ltd.