The fifth MLMI workshop will be held in Utrecht, The Netherlands, following successful workshops in Martigny (2004), Edinburgh (2005), Washington (2006) and Brno (2007). MLMI brings together researchers from the different communities working on the common theme of advanced machine learning algorithms applied to multimodal human-human and human-computer interaction. The motivation for creating this joint multi-disciplinary workshop arose from the actual needs of several large collaborative projects, in Europe and the United States.