In vitro plaque removal studies require biofilm models that resemble in vivo dental plaque. Here, we compare contact and non-contact removal of single and dual-species biofilms as well as of biofilms grown from human whole saliva in vitro using different biofilm models. Bacteria were adhered to a salivary pellicle for 2 h or grown after adhesion for 16 h, after which, their removal was evaluated. In a contact mode, no differences were observed between the manual, rotating, or sonic brushing; and removal was on average 39%, 84%, and 95% for Streptococcus mutans, Streptococcus oralis, and Actinomyces naeslundii, respectively, and 90% and 54% for the dual- and multi-species biofilms, respectively. However, in a non-contact mode, rotating and sonic brushes still removed considerable numbers of bacteria (24–40%), while the manual brush as a control (5–11%) did not. Single A. naeslundii and dual-species (A. naeslundii and S. oralis) biofilms were more difficult to remove after 16 h growth than after 2 h adhesion (on average, 62% and 93% for 16- and 2-h-old biofilms, respectively), while in contrast, biofilms grown from whole saliva were easier to remove (97% after 16 h and 54% after 2 h of growth). Considering the strong adhesion of dual-species biofilms and their easier more reproducible growth compared with biofilms grown from whole saliva, dual-species biofilms of A. naeslundii and S. oralis are suggested to be preferred for use in mechanical plaque removal studies in vitro.